Simultaneous time resolution of the emission spectra of fluorescent proteins and zooxanthellar chlorophyll in reef-building corals[para][dagger]

Photochemistry and Photobiology, May 2003 by Gilmore, Adam M, Larkum, Anthony W D, Salih, Anya, Itoh, Shigeru, Et al

The spectral kinetic model was robustly fit (27,28) using the robust statistical norm and curve-fitting (L^sub 1^) method by minimizing the sum of absolute deviations, where the absolute deviation for a given time-wavelength channel coordinate was L^sub i^ = |D^sub i^ - M^sub i^|, where D^sub i^ is the datum and M^sub i^ is the model prediction of the datum. The global analysis program was written for use in Microsoft Windows NT4.0 (32 bit) and Excel 97, with Visual Basic for Applications97 and uses a large-scale general reduced gradient (LSGRG) minimization engine developed by Frontline Systems Inc. (Incline Village, NV). The LSGRG engine is capable of handling 4000 parameters and 4000 constraints and uses a sparse matrix representation. The LSGRG engine is capable of solving, in Excel 97 with a Pentium III computer with 500 Mb RAM and an 800 MHz processor, all 26 test problems prescribed by the National Institute of Standards and Technology website for nonlinear regression (http://www.itl.nist.gov/div898/strd/nls/ nls_info.shtml) with 10 digit precision for the sum of squares parameter. Standard errors on fluorescence lifetime mode center and width parameters longer than the instrument response function (~50 ps) were estimated from repeated simulations to be less than 10%, whereas lifetime modes and widths below this value were estimated to have ambiguity factors approaching 2.

RESULTS

Simultaneous time resolution of FP and zooxanthellae PSII fluorescence emission spectra

Figure 2 illustrates a comprehensive view of the time and wavelength dependence of both the FP and zooxanthellae fluorescence from a green color morph specimen of P. versipora harvested from Sydney Harbor, Australia. Figure 2a illustrates the decay kinetics of the clearly separate FP and chl emission components. The FP fluorescence bands (peak ~515 nm) emitted with stronger amplitudes (left side of main panel) and slower decay times compared with the algae's chl fluorescence bands (peak 683 nm, upper right of main panel). Figure 2b profiles the data and model fits on a log scale to illustrate the slower main FP emission component in direct comparison with the more rapid PSII component. Figure 2c illustrates the same spectral data as in Fig. 2a,b plotted as a function of time, where each time channel is normalized to the peak FP emission wavelength intensity. It is clear that all components of the chl exhibit kinetics that are independent of the FP emission. Figure 2d shows the main fluorescence lifetime decay distribution components with positive amplitudes used to simulate the chl (red) and FP emission (green) in Fig. 2b. The major chl distribution mode was significantly broader and centered at a much faster lifetime value compared with the narrow FP emission component. The model simulations all clearly indicated that there were no kinetic components resolved in the analysis to show that the excited-state population of the FP exhibited any significant correlation with that of the algae's chl, i.e. except for the original laser excitation pulse event. Interestingly, however, the raw data normalization procedure in Fig. 2c did clearly reveal that the initial kinetic phase of the FP rise, during and immediately after the laser excitation (0-300 ps), is associated with a small re-producible greenshift in the fluorescence. We interpret this shift to be consistent with rapid resonance energy transfer (FRET) from blue to greener FP proteins. The evidence for FRET is examined in more detail later.