Identification of immobile single molecules using polarization-modulated asynchronous time delay and integration-mode scanning

Abstract

We report the development of a data acquisition method for identifying single molecules on large surfaces with simultaneous characterization of their absorption dipole. The method is based on a previously described device for microarray readout at single molecule sensitivity (Hesse, J.; Sonnleitner, M.; Sonnleitner, A.; Freudenthaler, G.; Jacak, J.; Höglinger, O.; Schindler, H.; Schütz, G. J. Anal. Chem. 2004, 76, 5960-5964). Here, we introduced asynchronous time delay and integration- (TDI-) mode imaging to record also the time course of fluorescence signals: the images thus contain both spatial and temporal information. We demonstrate the principle by modulating the signals via rotating excitation polarization, which allows for discriminating static absorption dipoles against multiple or freely rotating single absorption dipoles. Experiments on BSA carrying different numbers of fluorophores demonstrate the feasibility of the method. Protein species with an average labeling degree of 0.55 and 2.89 fluorophores per protein can be readily distinguished.

Figure 1

Fluorescence images of BSA_low (a-c) and BSA_high (d). (a) shows a synchronous TDI-mode image where the individual molecules are visible as diffraction limited peaks with high signal-to-noise ratio. (b) Asynchronous TDI-scan where individual diffraction limited objects are imaged as stretched peaks. Intensity fluctuations along the peaks can be attributed to photoblinking of the Atto647 dye molecules. (c+d) Polarization-modulated asynchronous TDI-mode images. While BSA_low molecules (c) show a strong fluorescence modulation and no signal between the maxima, BSA_high molecules (d) are characterized by a low modulation and significant fluorescence signal also between adjacent maxima. Images in (b) and (c+d) were acquired with v_sample = 5μm / s and v_sample = 4μm / s, respectively.

Figure 2

(a) Absorption dipole $\overrightarrow{d}$ and excitation vector $\overrightarrow{p}$ in the presented experimental configuration with θ characterizing the orientation of the dipole with respect to the xy-plane (elevation) and φ the angle between excitation vector and projection of the absorption dipole in the xy-plane (azimuth). (b) The emission intensity I is given by I = I₀ · cos² φ · cosθ with I₀ the intensity for parallel/anti-parallel orientation of $\overrightarrow{d}$ and $\overrightarrow{p}$. Readout parameter are the intensity of a local minimum I_min and the background signal I_bg. The offset for each stretched peak is calculated as $I_{valley}=\overline{I_{\min }}-\overline{I_{bg}}$.

Figure 3

Fluorescence traces of single BSA molecules acquired in polarization-modulated asynchronous TDI-mode scanning. The solid lines show traces of an individual BSA_low (a) and BSA_high (b) molecule; the dashed lines show the background signal next to the stretched peaks. The determined modulation frequency f_fl of ~4 Hz agrees well to the expected value for a 2 Hz polarizer frequency f_ex. The average modulation depth I_valley is clearly different for the two BSA species.

Figure 4

Probability analysis of the degree of labelling. (a) and (b) show the probability distributions of I_valley for BSA_low and BSA_high, respectively. In both cases, the probability density functions of measured data (black dots) agree well with Monte Carlo simulations (gray line). Note the pronounced tail towards higher valley intensities for strong labeled molecules. c) The degree of labeling was estimated by comparison with simulated data via the Kolmogorov-Smirnov test. Simulations were performed with different labeling degrees λ (Poissonian distribution of labeling). Experiments on BSA_low (gray line) and BSA_high (black line) were analyzed. The dotted horizontal line indicates the significance level of 0.05. Dashed vertical lines show the λ values calculated – based on the UV-VIS spectroscopy measurements – for the actually labeled fractions of $BS{A}_{low}^{app}=1.3$ (gray) and $BS{A}_{high}^{app}=3.1$ (black). The analysis reveals distinct maxima for the two differently labeled molecule species with maximum p-values of ~0.14 (λ = 0.98) and ~0.13 (λ = 2.38) for BSA_low and BSA_high, respectively.

Figure 5

Probability for observing molecules carrying 1 to 8 fluorophores when measuring a particular I_valley of 0 counts (circles), 10 counts (filled stars; experimentally determined value for BSA_low), 30 counts (squares), or 50 counts (triangles); I₀ was set to the measured peak value of 109 counts for BSA_low. For increasing values of I_valley the labeling distribution shifts towards higher numbers of dyes per molecule n reaching 0 % singles at I_valley = 50 counts.