Discussion

concentra-76 3. FCS, Confocal Volume, Concentration, Molecular Brightness and Artifacts

tions. While it is probably not possible to prevent surface adsorption for low concentrations, the accuracy of the sample preparation method can be improved e.g. by using gravimetrical dilution.

The other method to extract the effective volume from FCS measurements requires the knowledge of the diffusion coefficient of the fluorophore under investigation. There is no need to know the exact sample concentration and sample loss due to adsorption does not influence the result. The effective volume is extracted by fitting a model function to the correlated data, which also requires assumptions to be made about the MDF. The MDF was assumed to be a 3D Gaussian function. Since this approach involves fitting of the model function to the experimental data, the quality of the correlated data also influences the precision of the findings. It was found that the determination of the confocal volume was only possible for samples with concentrations between 100 pM and 5 nM. Altogether, the effective volume resulting from FCS fits can only be determined as precise as the used model describes the experimental reality. With this method the confocal volume was determined to be (1.15±0.06) fl.

By scanning a fluorescent microsphere an effective volume of (1.0±0.1) fl was found. Be-cause of optical saturation occurring in the FCS experiments, the effective volume determined with both FCS methods is expected to be larger by 13%. While for the fitting approach this expectation is satisfied, the dilution series yields the same effective volume as the imaging of the fluorescent microspheres. The uncertainties of all three methods however are rather large and the saturation effect might not be resolvable.

Regarding the uncertainties of the three methods no clear recommendation for the deter-mination of the effective volume can be given. However, the two methods employing FCS measurements have the advantage that the effective volume is measured in aqueous envi-ronment and distortions of the MDF due to this envienvi-ronment are taken into account. The dilution series approach offers additionally strong advantages over the two other methods since it does not depend on model assumptions.

To analyze the limits of the commonly used 3D Gaussian model for the MDF the influence of optical saturation, cover-slide thickness deviation, different pinhole diameters and different beam waists was investigated. In agreement with the calculations done by Enderlein et al. [64]

the measurements show that saturation significantly influences the MDF. The increase of the

3.6. Discussion 77

effective volume will lead to apparently higher concentrations. Even for modest excitation power (P ∼ 0.4·Psat) the confocal volume was found to increase by 13% compared to the extrapolated effective volume for no excitation. The increase of the effective volume due to saturation has to be accounted for, especially as FCS measurements have to be performed with high excitation intensities in order to obtain a sufficient count rate per molecule. The fact that saturation depends on the used fluorophore implements the necessity to analyze saturation characteristics for the fluorophore under investigation and to take it into account in the analysis. Without the occurrence of photo-bleaching, optical saturation results in an apparantly smaller diffusion coefficient or a larger confocal volume, depending which parameter is fitted. Evidence for photo-destruction of the fluorophores was found at higher laser power levels which leaded to lower diffusion times and ultimately could resulted in a wrongly determined effective volume or diffusion coefficient.

Due to optical saturation not only the size of the confocal volume but also its shape changes. The assumption of a 3D Gaussian shape, therefore, with increasing excitation in-tensities, leads to increasing systematical errors. In agreement with Enderlein et al. [64] I conclude, that saturation effects always need to be taken into account in FCS experiments and that fluorophores and excitation intensities need to be chosen carefully to avoid misin-terpretation of the findings.

The experiments showed that the molecular brightness depends on the diameter of the confocal pinhole. By decreasing the pinhole diameter, a higher confocality and thus a smaller confocal volume is achieved but this increase in resolution is paid by a decrease of the molecu-lar brightness. Decreasing the excitation beam width on the other hand increases the confocal volume but also decreases the molecular brightness since the local excitation intensity is re-duced. Therefore, excitation beam width and pinhole diameter have to be chosen according to each other.

The cover slide thickness correction ring setting of the objective is influencing the shape of the MDF. Besides the ability to adjusted the objective to different cover slides, it can be used to correct for refractive index mismatch and other aberrations. By imaging fluorescent microspheres for different correction ring settings the shape of the MDF can be monitored and optimized.

The MDF is sensible to the experimental conditions. It is therefore most desirable to

78 3. FCS, Confocal Volume, Concentration, Molecular Brightness and Artifacts

calibrate the effective volume under the exact same conditions that apply in the experiment of interest. For the same reason it is advisable to use the same fluorophores during the exper-iments that were used for calibration, since a change in the photo-physic of the fluorophore will yield a change of the MDF. If this is not possible saturation effects have to be quantified and considered in the analysis.

With respect to the experimental protocol of FCS experiments I suggest to measure the MDF with fluorescent microspheres as this techniques yields an image of the MDF and arti-facts can directly be identified. If the adjustment of the apparatus is found to be sufficient to be approximated by a 3D Gaussian function, then the confocal volume can be extracted from a fit of the correlation of the dye that will be used in the experiment, given its diffusion coefficient is known. Measuring the confocal volume through a dilution series is undoubtedly elaborate, but offers the advantage of being model independent and that the diffusion coef-ficient does not need to be known. The accuracy in this case only depends on the accuracy of the sample preparation. The effect of saturation on the MDF should always be taken into account, therefore the saturation curve should be recorded for all dyes employed in the determination of the effective volume and the final FCS measurements. To our finding a power of P = 0.4·Psat is the best compromise between low saturation and good signal to background ratio.

4. POLYPROLINE AS CALIBRATION ASSAY FOR FRET DISTANCE