Identification of different emitting species in the red fluorescent protein DsRed by means of ensemble and single-molecule spectroscopy

Abstract

The photophysics and photochemistry taking place in the DsRed protein, a recently cloned red fluorescent protein from a coral of the Discosoma genus, are investigated here by means of ensemble and single-molecule time-resolved detection and spectroscopic measurements. Ensemble time-resolved data reveal that 25% of the immature green chromophores are present in tetramers containing only this immature form. They are responsible for the weak fluorescence emitted at 500 nm. The remaining 75% of the immature green chromophores are involved in a fluorescence resonance energy transfer process to the red species. The combination of time-resolved detection with spectroscopy at the single-molecule level reveals, on 543-nm excitation of individual DsRed tetramers, the existence of a photoconversion of the red chromophore emitting at 583 nm and decaying with a 3.2-ns time constant into a super red one emitting at 595 nm and for which the decay time constant ranges between 2.7 and 1.5 ns. The phenomenon is further corroborated at the ensemble level by the observation of the creation of a super red form and a blue absorbing species on irradiation with 532-nm pulsed light at high excitation power. Furthermore, single-molecule experiments suggest that a similar photoconversion process might occur in the immature green species on 488-nm excitation.

In recent years, green fluorescent protein (GFP) has become one of the most popular research tools in cell and molecular biology, being used as a noninvasive fluorescent marker for gene expression, protein localization in living cells, or in protein–protein interaction studies via fluorescence resonance energy transfer (FRET) (1–3). Via point mutations, variants were developed having emission colors from blue to yellowish green, increased brightness, and photostability (1). Isolated or fused to nonfluorescent proteins, GFPs can be monitored easily by means of fluorescence microscopy. Moreover, pairs of GFP mutants having different emission properties can form donor–acceptor systems for FRET or can allow simultaneous localization of two or more fusion proteins in multicolor tracking applications (4, 5). Because fluorescence microscopy performed on living cells requires labeling with fluorophores having long-wavelength absorption/emission properties to avoid autofluorescence, much effort was expended in creating GFP variants with absorption/emission shifted to the red. The most red-shifted mutant of the wild-type GFP, the Enhanced Yellow Fluorescent Protein, has its emission maximum at 529 nm.

Recently, the cloning and expressing in vivo of a new highly fluorescent red protein (DsRed) from a coral of the Discosoma genus was demonstrated (6). Having a structure similar to GFP but with the emission maximum shifted to 583 nm, DsRed appears to be a longer-wavelength substitute for GFPs. Moreover, mutants of DsRed with the emission maximum shifted even more to the red were already reported (7–9). Previous publications revealed a part of the biochemistry and photophysics involved in the DsRed protein and the suitability of this protein for in vivo applications (7–19). Several findings pointed toward some major drawbacks for DsRed when used as a fluorescent probe. Analytical ultracentrifugation, time-resolved anisotropy, and single-molecule spectroscopy (SMS) experiments revealed the occurrence of DsRed as a tetramer, even at subnanomolar concentration (7, 10, 14). The recently solved crystallographic structure confirmed the tetramerization and demonstrated the arrangement of the chromophores within the tetramer in antiparallel pairs (12, 17). Moreover, the final red-emitting chromophore appears through a slow and incomplete maturation process, resulting in tetramers containing one or more green precursor chromophores even for so-called mature DsRed samples. As a result, FRET can occur between the above-mentioned partners. Relying on the crystallographic data, the rate constants for energy hopping between the chromophores within a tetramer were calculated (16). Collective effects resulting from energy transfer between the chromophores within a tetramer were observed by means of SMS experiments (14).

In the present paper, we take a closer look at the photophysical behavior of DsRed by means of detailed ensemble and single-molecule time-resolved and spectroscopic studies.

Experimental Procedure

Purification of the Wild-Type DsRed Protein.

Details about the expression and purification procedures for DsRed were published elsewhere (14). The proteins were stored in a PBS solution (10 mM, pH 7.4, Sigma).

Ensemble Spectroscopy and Photoconversion.

Steady-state spectroscopy was performed on a Perkin–Elmer Lambda 40 spectrophotometer and a Spex Fluorolog 1500 fluorimeter (Spex Industries, Metuchen, NJ). Picosecond laser-induced time-correlated single-photon counting (TCSPC) fluorescence spectroscopy on excitation at 460, 543, and 578 nm was performed on a 10⁻⁷ M DsRed solution in PBS. The fluorescence decays were collected in 4,096 channels by using a time increment of 6 ps and globally analyzed over the emission spectrum by convoluting the experimental instrumental response function of the TCSPC set-up (full width half maximum of 30 ps) with an exponential model function containing a time shift and a baseline for uncorrelated background. Depending on their contribution to the total detected fluorescence, decay times as short as 5 ps can be determined. In addition, time-resolved emission spectra on excitation at 460 nm were recorded from 470 to 700 nm, with a 10-nm step by time-gating the single-photon timing detection system and scanning the emission spectrum.

A 10⁻⁵ M DsRed solution in PBS, separated into two quartz cells of 3-mm path, was irradiated for about 14 h with the 532-nm light provided by a nanosecond pulsed frequency-doubled Nd-YAG laser (Spectra-Physics) at an average integral power at the sample of 120 mW (1.5-MW peak power) and 200 mW (2.5-MW peak power), respectively. During the irradiation experiment, absorption and fluorescence spectra were recorded from both solutions. For the final photoconverted DsRed solutions, time-resolved fluorescence decays on 578-nm excitation were recorded and analyzed as mentioned before.

Single-Molecule Detection and Spectroscopy.

Samples for SMS experiments were prepared by immobilizing individual DsRed tetramers in either water-filled pores of polyvinylalcohol (PVA; see ref. 14) or polyacrylamide (PAA) gel (see ref. 20).

Picosecond time-resolved confocal fluorescence detection and spectroscopy of individual DsRed tetramers were performed by using an optical setup that is described elsewhere (21). Briefly, samples were mounted on an Olympus IX 70 inverted microscope equipped with a scanning stage (Physics Instruments, Waldbrown, Germany). Excitation with either 488- or 543-nm pulsed laser lines occurred by an oil immersion objective (Olympus 1.4 N.A., ×60, Olympus). Fluorescence was collected by the same objective, passed through dichroic mirrors (Chroma Technology, Brattleboro, NY), filtered through a notch filter (Kaiser Optical Systems, Ann Arbor, MI), and simultaneously focused, via a 50-μm pinhole and a 50–50% nonpolarizing beam-splitter cube, on an avalanche photodiode (APD) (SPCM 15, EG & G, Quebec) and into the entrance of a polychromator (Acton SP 150) coupled to a cooled charge-coupled device camera (Princeton Instruments). After parking an individual DsRed spot in the laser focus, the fluorescence signal detected by the APD was registered by a time-correlated single-photon counting personal computer card (SPC 630, Picoquant, Berlin) by using the burst-integrated fluorescence lifetime (BIFL) technique (22). By using pulsed excitation and BIFL detection, for each detected photon, the time position with respect to the excitation pulse and the time lag with respect to the previous detected photon are stored. The BIFL data set allows the reconstruction of the fluorescence time trace (transient) with a dwell time of at least 50 ns, as well as of the time-resolved fluorescence decays by using photons belonging to particular regions of the transient, i.e., regions of constant intensity. In this way, several decay histograms stored in 256 channels with a resolution of 100 ps/channel, and at least 100 counts in the peak were obtained for each individual DsRed tetramer. The reconstructed fluorescence decays were analyzed by a maximum likelihood estimator by convoluting the experimental instrumental response function (full width half maximum of 400 ps) with an exponential model function containing a time shift, a baseline for uncorrelated background, and a scaling factor for the correlated background (22, 23). In addition, fluorescence spectra and transients from individual DsRed tetramers were recorded by using either 488- or 543-nm continuous wave (CW) excitation.

Results

Ensemble Time-Resolved Spectroscopy.

As previously reported (6, 7), the absorption spectrum of DsRed displays a strong peak located at 559 nm and two additional shoulders at 524 and 485 nm. Independent of the excitation wavelength, fluorescence is emitted with a major peak at 583 nm. An additional peak, weak in intensity and located at around 500 nm, is present when DsRed is excited at 450 nm (15). Although the 559-nm absorption and 583-nm emission peaks were assigned to the red chromophore within the tetramer, the absorption shoulder at 485 nm together with the emission peak at 500 nm were attributed to an immature green precursor form of the red chromophore (6, 7). Depending on the excitation wavelength, the fluorescence emitted by DsRed displays either mono- or multiexponentially temporal behavior (see below). When the protein is excited in the absorption band of the red form at either 543 or 578 nm, the fluorescence decays, independent of the monitoring wavelength (from 560 to 670 nm on 543-nm excitation and from 590 to 670 nm on 578-nm excitation), monoexponentially with a time constant of 3.60 ns, in agreement with previous reports (10, 16). By contrast, direct excitation in the absorption band of the immature green form at 460 nm leads to a multiexponential decay of the fluorescence over the whole emission spectrum of the protein, i.e., from 500 to 650 nm. Four components, having the values of 3.60, 0.47, 0.12, and 0.03 ns, were necessary to globally fit the decays measured over the emission spectrum from 500 to 650 nm. Their contribution to the total detected fluorescence is depicted in Fig. 1a. The 3.60- and 0.47-ns components contribute only positively to the detected fluorescence. The 3.60-ns component dominates the fluorescence detected above 560 nm, whereas its contribution decreases at shorter wavelengths and becomes zero at 500 nm. The 0.47-ns component is present only in the emission region of the immature green chromophore, with a maximum contribution of 3% at 500 nm (see Fig. 1a Inset). The fluorescence detected at 500 nm is mainly built up by the 0.12- and 0.03-ns components, their amplitudes having values of 16 and 80%, respectively (see Fig. 1 a and c). These two fast components are observed as rise terms in the fluorescence decays detected above 560 nm (see Fig. 1 a and d). However, their relative amplitudes detected at 650 nm (−8 and −60% for the 0.12- and 0.03-ns components, respectively) do not match the values detected at 500 nm. Time-resolved emission spectra detected on 460-nm excitation show, apart from the instantaneous appearance of the 583-nm band, the disappearance of the 500-nm peak (see Fig. 1b).

Figure 1.

(a) Wavelength dependence of the weighted amplitudes corresponding to the 0.03- (line + star), 0.12- (line + triangle), 0.47- (line + circle), and 3.60-ns (line + square) fluorescence decay times of DsRed in PBS buffer excited at 460 nm. (Inset) Zoom of the 0.47-ns component. (b) Time-resolved emission spectra of DsRed in PBS excited at 460 nm. Fits and residuals of the fluorescence decays detected at 500 nm (c) and 650 nm (d) on 460-nm excitation.

Photoconversion.

When irradiated in the absorption band of the red chromophore with 532-nm pulsed laser light by using 120-mW excitation power at the sample, DsRed dramatically changed its absorption and fluorescence properties. On irradiation, the absorption peak at 558 nm diminished and shifted bathochromically up to 574 nm, and a new band appeared at 386 nm (see Fig. 2a). Interestingly, the absorption band associated with the immature green chromophore (peak at 479 nm) modified neither in position nor in intensity (see below). Two isobestic points, at 343 and 467 nm, could be observed, indicating a link between the disappearance of the 558-nm peak and the appearance of the two new bands at 386 and 574 nm. Relying on the fact that the absorption band of the immature green chromophore remained constant over the time of the irradiation with 532-nm laser light, the absorption spectra of the DsRed protein could be decomposed by using Gaussian functions into three bands (see Fig. 3 a and b): band B peaking at around 386 nm, band G peaking at around 486 nm, and a third band whose peak shifts as a function of the irradiation time from 558 (band R in Fig. 3a) to 574 nm (band SR in Fig. 3b). The time evolution of the relative peak absorbances of the three decomposed bands (the value of the peak absorbance of a band weighted to the sum of the peak absorbances of all three bands) on irradiation with 532-nm laser light is depicted in Fig. 2b. Although the relative peak absorbance of band G remains constant in time, for the other two bands, a correlation can be observed. Indeed, linear fits to the time evolutions of the relative absorbances of the B and R bands resulted in straight lines with slopes of 0.6 and −0.55 min⁻¹, respectively, confirming the existence of a photoconversion involving the forms connected to the R and B bands. As mentioned before, independently of the excitation wavelength (either 390, 460, or 570 nm), the fresh DsRed sample emits fluorescence with a major peak at 583 nm (see Fig. 3c). In contrast, the fluorescence emitted by the photoconverted DsRed strongly depends on the excitation wavelength (see Fig. 3d). On 390-nm excitation, fluorescence is emitted with a major peak at 500 nm and two pronounced shoulders at around 450 and 600 nm, whereas excitation at 460 nm gives rise to fluorescence peaking at 500 nm. Red edge excitation of the photoconverted DsRed at 570 nm gives rise to a new fluorescence band peaking at around 595 nm (see Fig. 3d). After resting in the dark at 277 K for 24 h, no modifications either in the absorption or fluorescence spectral properties were observed for the photoconverted DsRed solution. Similar spectral modifications were observed for the DsRed solution irradiated with 532-nm pulsed laser light at an excitation power of 200 mW.

Figure 2.

(a) Time evolution of the absorption spectrum of DsRed in PBS on irradiation with 532-nm pulsed laser light at an excitation power of 120 mW. (b) Time evolution of the relative peak absorbances of the R (squares), G (circles), and B (stars) bands (see text) during the irradiation experiment together with the corresponding linear fits (dash lines).

Figure 3.

(a) Absorption spectrum of fresh DsRed in PBS together with the decomposed R and G bands. (b) Absorption spectrum of the final photoconverted DsRed solution together with the decomposed SR, G, and B bands. (c) Fluorescence spectra of DsRed in PBS excited at 390 (dash line), 460 (line + squares), and 570 nm (line + circles). (d) Fluorescence spectra of the photoconverted DsRed solution excited at 390 (dash line), 460 (line + squares), and 570 nm (line + circles).

Ensemble Time-Resolved Spectroscopy on Photoconverted DsRed Protein.

The fluorescence decay detected at 670 nm on excitation of the photoconverted DsRed at 578 nm displays a multiexponential behavior. The recovered decay time parameters depend on the excitation power used in the photoconversion experiment. By using a three exponential model function, the main contribution comes from the long decay times with values of 2.78 ns for the sample photoconverted with 120 mW excitation power and 1.72 ns for the sample photoconverted with 200-mW excitation power.

SMS.

By using either 488- or 543-nm pulsed excitation, we performed simultaneous acquisition of fluorescence transients, time-resolved decays, and spectra from individual DsRed proteins immobilized either in PVA or PAA matrix. In addition, fluorescence transients and spectra of individual DsRed tetramers were recorded on excitation at either 488 or 543 nm in the CW regime.

Fluorescence Transients on CW Excitation at 543 and 488 nm.

The general trend previously observed for the transients recorded from individual DsRed tetramers in PVA on excitation with 543-nm CW laser light at low excitation power (0.5 kW/cm² (14)] was found back for both 488- and 543-nm CW excitation and for both PVA and PAA matrices, i.e., different intensity levels (from four to one), jumps between different levels as well as to the off level from every intensity level. It is interesting to note that for excitation powers having values below 3 kW/cm², the number of emitted photons per bin time increases with the increase of the excitation power. By applying higher excitation powers, the transients display, in the highest intensity level, count rates similar to those recorded with 10 times lower excitation power, indicating that the molecules are driven in saturation (results not shown), similar to what was previously found for a GFP mutant (24).

Fluorescence Decays and Spectra on Pulsed Excitation at 543 nm.

Typical transients detected from individual DsRed tetramers in PAA on excitation with 543-nm pulsed light are depicted in Fig. 4 a and b. However, as compared with the CW excitation regime (14), for the same average integral power of 0.5 kW/cm², the survival time as well as the intensity of the highest level within the transients detected on pulsed excitation are lower, suggesting that individual DsRed tetramers are driven in saturation already at this excitation power when pulsed excitation is used.

Figure 4.

(a and b) Fluorescence transients and corresponding decay times obtained on excitation of individual DsRed tetramers immobilized in PAA with 543-nm pulsed light. Histograms and Gaussian fits (line + square) of the decay times obtained on excitation at 543 nm of individual DsRed tetramers immobilized in PAA (c) and PVA (d). Histograms and Gaussian fits (line + square) of the emission maxima of the fluorescence spectra recorded on excitation with 543-nm pulsed light of individual DsRed tetramers immobilized in PAA (e) and PVA (f).

The fluorescence decays reconstructed from photons belonging to constant intensity levels within the transients could be fitted monoexponentially, independent of the excitation wavelength and matrix used (Fig. 4 a and b). The distributions of the recovered decay times on 543-nm excitation are depicted in Fig. 4c for PAA and Fig. 4d for PVA. They contain data from about 65 individual tetramers for each matrix. Gaussian fits of the distributions yielded peak values of 1.42 and 2.71 ns for PAA and 1.56 and 2.64 ns for PVA, respectively. Because the maximum likelihood estimator analysis can deliver, for decays containing more than 1,000 total detected counts, values with theoretical standard deviation of at most 150 ps (23), the two values recovered from individual DsRed tetramers embedded both in PAA and PVA can be considered as being similar. However, their relative contribution changes from PAA to PVA (Fig. 4 c and d). The long decay times can be recovered in all of the intensity levels, whereas the short ones appears mainly in the low intensity levels. Interestingly, the decay times recovered from the highest levels in the beginning of the transients, especially in the case of PVA, display values larger than 3 ns, their histogram peaking at around 3.2 ns (data not shown). As an example, two typical transients together with the recovered decay times from specific regions are depicted in Fig. 4 a and b. The distributions of the emission maxima obtained from the analysis of the fluorescence spectra recorded on 543-nm pulse excitation of individual DsRed tetramers are depicted in Fig. 4 e and f. Gaussian fits to these monomodal distributions yielded peaks at 595 nm for PAA and 594 nm for PVA. During the spectral runs of individual DsRed tetramers, shifts of the emission maximum from 583 to 600 nm were observed.

Fluorescence Decays and Spectra on Pulsed Excitation at 488 nm.

In contrast to the 543-nm pulsed experiment, excitation into the band of the immature green chromophore leads to a more complicated distribution. Apart from the two values found on 543-nm excitation, i.e., 1.5 and 2.7 ns, an additional fast component of 0.9 ns in PAA and 1.1 ns in PVA was recovered, only from the low emitting levels within the transients (data not shown). However, for individual DsRed tetramers immobilized in PVA, excitation at 488 nm leads to an increased contribution of the decay times larger than 3 ns, whereas in PAA, these values are almost absent.

On 488-nm excitation, the emission maxima of the fluorescence spectra detected from individual DsRed tetramers, depending on the matrix, group into either a bimodal distribution for PAA (peaks at 547 and 595 nm) or a trimodal one for PVA (549, 584, and 593 nm). The green form, emitting at around 547 nm in PAA and at 549 nm in PVA, was present only in 30% of the investigated DsRed tetramers (90 individual tetramers) within their spectral run, alone or together with one of the red bands.

Because the decay times and emission maxima were recovered from the analysis of the simultaneously recorded fluorescence decays and spectra from individual DsRed tetramers, a connection between these values was observed (see below). For PVA, the fluorescence spectra peaking at around 584 nm were present together with a 3.2-ns decay time (see Fig. 5 a and d), whereas for those peaking at 593 nm, decay times of around 2.7 and 1.5 ns were observed (see Fig. 5 b and e). The fluorescence spectrum peaking at around 547 nm, when present alone, was connected with a decay time of 3 ns (see Fig. 5 c and f). Similar observations were found for PAA, except for the fluorescence spectra peaking at 584 nm.

Figure 5.

Single-molecule fluorescence spectra of (a) red, (b) super red, and (c) immature green forms identified on excitation at 488 nm of individual DsRed tetramers immobilized in PVA. Fluorescence decays together with the corresponding fits corresponding to the red (d), super red (e), and green spectral (f) forms observed on excitation at 488-nm wavelength of an individual DsRed tetramers in PVA.

Discussion

FRET from the immature green to the red chromophore within the DsRed tetramer was already suggested, especially on the basis of the spectral overlap existing between the emission/absorption of these two forms (7, 15). Moreover, the existence of tetramers containing from none to four green immature chromophores was recently demonstrated by two-color excitation single-molecule experiments (25). Our ensemble time-resolved data clearly prove the existence of FRET within the DsRed tetramers (see below). Although direct excitation in the absorption band of the red chromophore (at either 543 or 578 nm) leads to emission only from the red chromophore, the fluorescence detected above 590 nm decaying monoexponentially with a time constant of 3.60 ns, excitation in the absorption band of the immature green chromophore (460 nm) results in a complicated excited state dynamics (see Fig. 1a). Emission from the red chromophore is detected on excitation outside its absorption band (see Fig. 1 a and b) and appears, as shown by the time-resolved emission spectra, almost instantaneously. Fluorescence from the green chromophore, peaking at 500 nm, is quenched, i.e., by FRET, two fast components of 0.026 and 0.12 ns being detected as decay terms in the green and as rise terms in the red emission part (see Fig. 1). In our opinion, FRET from the green chromophore to a red partner within the tetramer will take place, depending on the orientation between the partners, in 0.03 ns to the most properly oriented chromophore and in 0.12 ns to a chromophore having a less proper orientation. This is in agreement with the recently calculations of Moerner and coworkers (16), which, on the basis of crystallographic data (17), suggested different rate constants of energy hopping between the four red chromophores within a tetramer. Depending on the orientation of the chromophores, they obtained values between 0.01 and 17 ns. Similar orientation considerations should be taken into account for a unidirectional energy transfer from the immature green to a red chromophore. However, not all of the green forms absorbing at 460 nm undergo FRET, because the contribution of the 0.03- and 0.12-ns decay times detected at 500 nm do not match the contribution of the corresponding rise times at longer wavelengths, i.e., 650 nm (see Fig. 1a). As suggested by our TCSPC data, 25% of the green chromophores absorbing at 460 nm will decay radiatively with a time constant of 0.47 ns, and they can be seen as an emission peak at 500 nm in the time-resolved emission spectra. Hence, we attribute the 0.47-ns decay time to tetramers containing only green chromophores. The existence of the “green tetramers” was recently proven by SMS experiments (25). However, because from a chromophore with a structure similar to wild-type GFP, one would expect a decay time of around 3 ns, a decrease to 0.47 ns would, as previously suggested, account for some GFP variants (26) for a chromophore that is not rigidly encapsulated in the barrel of the protein and for which additional nonradiative channels might be opened. FRET is supported also by single-molecule data, because direct excitation of individual DsRed tetramers into the absorption band of the green chromophore (at 488 nm) leads mainly to the emission in the red spectral region.

When pulsed excitation is used, the photophysical processes in the DsRed protein become even more complicated. The amount of emitted photons per time unit, as well as the survival time of individual DsRed tetramers, decreases considerably as compared with CW excitation by using similar average power. A similar effect was observed on increasing the CW excitation power (data not shown) and hence is attributed to saturation. Furthermore, apart from saturation, the photoconversion of the red chromophore into a super red form was observed in the individual tetramers. As a consequence, the data obtained from the single-molecule experiments by using 543-nm pulsed excitation no longer display the behavior of a single emitting species in the red spectral region, i.e., monomodal distributions for the decay time and for the fluorescence maximum close to the ensemble values. Time-resolved single-molecule experiments reveal the presence of a bimodal distribution of decay times peaking at 1.5 and 2.7 ns for both PAA (see Fig. 4c) and PVA (see Fig. 4d), with an increased occurrence of the short decay time in the case of PAA. The spectral shifts observed during the recording of the fluorescence spectra from individual DsRed tetramers, as well as the broad distributions observed for the fluorescence maxima in both PVA (Fig. 4e) and PAA (see Fig. 4f), further support the hypothesis of photoconversion. Following the time evolution of the decay times as well as of the fluorescence maxima during irradiation, a part of the investigated tetramers shows in the beginning values of 3.2 ns and 583 nm (see Fig. 5 a and d), which fit well the ensemble data belonging to the red chromophore. On prolonged irradiation, a spectral form, i.e., the super red one, decaying within either 2.7 or 1.5 ns and emitting at around 595 nm, was observed; the fluorescence spectrum and decay associated with this form are depicted in Fig. 5 b and e. The creation of a super red chromophore in solution is clearly demonstrated by the time evolution of the absorption spectra during the ensemble irradiation experiment (see Fig. 2). With the absorption bathochromically shifted to 574 nm (band SR in Fig. 3c), as compared with the red chromophore (band R in Fig. 3a), the super red species emits, on excitation at 570 nm, at around 595 nm (see Fig. 3d). Moreover, ensemble TCSPC experiments performed on the photoconverted DsRed reveal the presence of a long decay time whose value decreased from 2.78 to 1.71 ns when the excitation power used during the irradiation was almost doubled. The coincidence between the values recovered from the ensemble photoconversion experiment and single-molecule measurements clearly demonstrate the creation of a super red species within the individual tetramers when pulsed excitation is impinged. On the basis of the above-mentioned experimental facts, we propose that irradiation of the red chromophore emitting at 583 nm will lead to the formation of a super red species emitting at 595 nm and decaying with a 1.5-ns time constant. The 2.7-ns decay time value can account for two possible situations: an intermediate species to the final super red form in the photoconversion process or a superposition of the fluorescence emitted from both the newly created super red and the red chromophores within an individual tetramer. Because at the single-molecule level the 2.7-ns decay time appears together with emission at 595 nm, and because the creation of either the long- or short-living super red species in solution is excitation power dependent, the first assumption seems to be more plausible.§ Moreover, creation of a super red species within a tetramer in which the red chromophores are initially coupled via energy hopping will lead to emission only from the super red form, i.e., at 595 nm, as it has the lowest energetically excited state. For a part of the investigated tetramers, mainly in PAA, a decay time of 3.2 ns together with an emission peaking at around 583 nm was not observed, indicating that photoconversion took place already during the imaging of the sample. Note that a photoconversion to a red-emitting species was previously reported for GFP when high-excitation blue laser light is applied on the protein under conditions of low oxygen concentration (27).

On excitation with 488-nm pulsed or CW laser light, part of the tetramers show green fluorescence peaking at 545 nm with a corresponding decay time of 3 ns (see Fig. 5 c and e). The reason why the green emission is shifted about 45 nm bathochromically in the SMS experiments (see Fig. 5 c and d) as compared with the ensemble spectrum reported here is not fully understood at this stage. One may speculate that a photoconversion process similar to the one observed for the red chromophore on 543 nm excitation might take place for the green chromophore(s) within the DsRed tetramers.

As mentioned in Results, together with the super red chromophore, a blue form absorbing at 386 nm and emitting at 450 nm appears from the photoconversion of the red form of DsRed (see Fig. 2; Fig. 3 b and d). As suggested by the mirror-like evolution of the relative absorbances of the blue and red band as a function of the irradiation time (see Fig. 2b), one could suppose that the blue form is nothing else than a protonated form. (A similar photoconversion to a blue-emitting species was reported for GFP mutants; see refs. 28 and 29). Ensemble TCSPC experiments performed on the photoconverted DsRed protein on 386- nm excitation reveal the presence of a fast component of about 25 ps with large contribution as a decay at 450 nm and a corresponding rise at 600, suggesting an excited-state proton transfer like that observed for wild-type GFP (data not shown) (30, 31).

The single-molecule data presented here point to the fact that the photoconversion of individual DsRed tetramers is accelerated in PAA as compared with PVA. Two facts can account for this difference: the difference in oxygen permeability between both matrices, meaning that oxygen plays a role in photoconversion, or the more rigid encapsulation of the individual DsRed tetramers in PVA as compared with PAA, meaning that an increase of the free volume for the DsRed tetramers when immobilized in PAA can accelerate the photoconversion process.

Conclusion

The present study reveals, by means of ensemble and single-molecule time-resolved and spectroscopic measurements, aspects related to the photophysical properties of DsRed. Ensemble time-resolved data reveal that 25% of the immature green chromophores are present in tetramers containing only this immature form. They are responsible for the weak fluorescence emitted at 500 nm. The remaining 75% of the immature green chromophores are involved in a FRET process to the red species. The combination of time-resolved detection with spectroscopy at the single-molecule level reveals the existence of a photoconversion of the red chromophore emitting at 583 nm and decaying with a 3.2-ns time constant into a super red one emitting at 595 nm and decaying either with either a 2.7- or a 1.5-ns time constant. The phenomenon is demonstrated at the ensemble level by the creation of the super red form on irradiation at high excitation power. Moreover, prolonged irradiation leads to the appearance of a blue form absorbing at 386 nm and emitting at 450 nm, possibly a protonated form of the super red one. Although at the single-molecule level the formation of the super red chromophore seems to take place through an intermediate species emitting at the same wavelength but decaying radiatively with a different time constant (2.7 ns), at the ensemble level, creation of one of these forms is excitation power dependent. On 488-nm excitation, the fluorescence detected from green chromophore(s) within individual DsRed tetramers decays with about 3 ns at 545 nm, different from the ensemble experiments. This might indicate a photoconversion process for the green chromophore(s) similar to that observed for the red one(s).

Our results point to the fact that, apart from the tetramerization and slow maturation as important drawbacks for DsRed when used as a fluorescent probe, photoconversion of the red chromophore to the super red one has to be taken into account when high laser excitation is used. Although these pitfalls seem to restrain the applicability of DsRed as a fluorescent marker, some groups have exploited them. Terskikh et al. suggested the use of slow maturation, i.e., the change of the fluorescence of DsRed from green to red over time as a fluorescent timer inside cells (8), whereas Marchant et al. proposed the use of the photoconversion/photobleaching process in the tetramers as a way to optically mark individual cells, organelles of fusion proteins (18). For quantitative applications (such as FRET), development of monomeric mutants with increased stability would be an important step forward for the use of DsRed proteins as fluorescent probes.

Abbreviations

GFP

green fluorescent protein

FRET

fluorescence resonance energy transfer

SMS

single-molecule spectroscopy

CW

continuous wave

PAA

polyacrylamide

PVA

polyvinylalcohol

TCSPC

time-correlated single-photon counting