EWDLS with a end-grafted DNA carpet

The first experiment with 2kbp DNA (Rg ∼100 nm) was done so that DNA was labelled with biotin and the surface chemistry was the same as already presented in chapter 2.2.2 (APTES silanized slides functionalized with glutaraldehyde and streptavidin). The goal of the measurement was to show (i) that we get a signal and that (ii) the measured signal is really coming from DNA. The latter was

(a) (b)

(c)

Figure 6.5: a)EWDLS intensity correlation functionsg⁽²⁾(τ)with114nm colloids (diameter) at 30^◦ with penetration depths of 400 nm and 600 nm. The pene-tration depth were obtained in the previous section as a result of the calibration with the fluorescent dye. The fits are done directly to the intensity correlation functions by using the equation (6.26) with the parameters describing the PMT and the single-mode fiber (see Appendix D). As a model for field autocorrelation function in equation (6.26) we used a single exponential (blue line) or the equa-tion (6.28) (red line). The free parameters in the fits were the diffusion constant D (see Table 6.1) and the local oscillator jl describing the scattering from the surface. The effect of increasing j_lis seen in the decreasing intercept of the inten-sity correlation function. This is because as the scattering volume gets smaller gets also the dynamic part of the signal smaller whereas the surface scattering stays roughly constant. The error bars are the standard deviation of the aver-aged measurements (50measurements for 30 seconds per angle and penetration depth). b) The residuals of the fits in (a). c) The residuals for the upper curve in (a) with penetration depths of 600 nm, 800 nm and 1000 nm. Furthermore we have also added the residuals of the fit with single exponential. The fits were performed as already explained in (a).

Table 6.1: Diffusion constants obtained from the best fits. The errors are es-timated by varying the measurement angle ±1^◦ and the penetration depth ±50 nm in the fits. The diffusion constants were not corrected for polydispersity.

θ[^◦] D [10⁻⁸cm²/s]

proven by addition of DNase I in a DNase buffer¹⁵ to the sample cell. DNase I is an endonuclease¹⁶ that cleaves DNA in small pieces. So that after addition of DNase I all DNA should be cut in small pieces and the measured signal should vanish since the small DNA pieces are expected give smaller scattering signal as coils at a fixed angle. Furthermore the small DNA pieces are free to diffuse away from the scattering volume which is not the case for end-grafted DNA. This should also reduce the scattering signal.

The first part of the measurement, where the dynamics of the DNA carpet was measured, was done in three angles θ = 35^◦,50^◦ and 75^◦. After that we measured the effect of DNase I addition to the DNA carpet at angle 50^◦. All the measurements were done at the maximum penetration depth of∼600 nm and in V-V geometry. Typical count rates in the measurements varied between 10 kHz forθ = 35^◦ and 4 kHz forθ = 75^◦. The results are shown in Fig. 6.6 where in (a) the measured intensity correlation functions g⁽²⁾(τ) are shown. The correlation functions before the addition of DNase I give a clear signal with faster decay at larger angles and after addition of DNase I the signal has vanished as discussed above. As was already seen in the measurement with the colloids some noise is inherently coupled to the measurements. Most of the noise problems in the measurements arose from the slow intensity variations which were in the range of few seconds. These variations distort the baseline of the correlation function.

This effect is even worse when the signal levels are low as is the case with DNA and therefore we had to apply a special filtering when we averaged the measured correlation functions. The filtering was done so that each single measurement run (30 seconds) was analyzed by looking at the last 15 time channels (1.888 s - 6.292 s) and then throwing out the ones where the sum of differences to the baseline was bigger than a user defined threshold. The user defined threshold

15In order to work DNase I needs a suitable buffer: 10mM Tris-HCl, 2.5mM MgCl2, 0.5mM CaCl2in pH 7.6.

16Endonuclease is a group name for enzymes (proteins) that cleaves its nucleic acid substrate at internal sites in the nucleotide sequence [81].

(a) (b)

10⁻⁴ 10⁻³ 10⁻² 10⁻¹ 10⁰ 10¹

0 0.2 0.4

Amplitude

10⁻⁴ 10⁻³ 10⁻² 10⁻¹ 10⁰ 10¹

0 0.2 0.4

10⁻⁴ 10⁻³ 10⁻² 10⁻¹ 10⁰ 10¹

0 0.2 0.4

Relaxation time [s]

35°

50°

75°

(c)

Figure 6.6: a)EWDLS measurement on end-grafted2kb DNA carpet cross-linked over biotin-streptavidin linkage to the surface. First three measurement were done at angles 35^◦, 50^◦ and 75^◦ following the addition of DNase I. The effect of addition of DNase I was measured at 50^◦ on two locations at the sample surface and all the measurements were done with the maximum penetration depth of

∼600 nm in V-V geometry. The DNase I cleaves DNA into small pieces causing the scattering signal to vanish as is also seen in the data. The error bars are the standard deviation of the averaged correlation functions. b) Normalized correlation functions measured before addition of DNase I and analyzed with S. Provencher´s CONTIN. The fits are calculated by equation 6.31 with the relaxation times and amplitudes presented in (c). c) The relaxation times and the amplitudes of the fits in (b).

was determined by observing what was the value for qualitatively good looking correlation function – that is a correlation function with a reasonable baseline.

After this filtering the averaged correlation functions constitute out of some 70− 90 correlation functions and measurements with DNase I have both 50 correlation functions. In Fig. 6.6 (a) the error bars are calculated after filtering and present the standard deviation of the averaged correlation functions. In Fig. 6.6 (b) the normalized correlation functions g⁽¹⁾(τ) are presented. Here the normalization was done with equation (6.18) by substituting j_l = 1−j_s and finding the value forjs so that the interceptg⁽²⁾(τ)−1 goes to 1. The error bars were also scaled with the correlation function. As the intercept of theg⁽²⁾(τ) was taken the average of 10−25 first time channels since the first time points are rather noisy and only through averaging we get a reasonable value for the intercept. The correlation functions measured before addition of DNase I were additionally analyzed with S.

Provencher´s CONTIN, which solves the inverse Laplace transformation¹⁷ of the correlation functions giving out amplitudes and relaxation times. The amplitudes and relaxation times of the correlation functions are then presented in Fig. 6.6 (c). The yellow fit curves in (b) are reconstructed by using equation

g⁽¹⁾(q, τ) = XN

i=1

A_ie^−t/τi, (6.31)

where Ai is the amplitude of the ith mode andτi is corresponding relaxation time found by CONTIN. The correlation functions constitute out of several decays so it is hard to analyze the data further but it is clearly visible that the relaxation times move to faster time scales as scattering vector gets larger as is expected for a diffusive behavior. Furthermore the origin of the signal should not be number fluctuations since for number fluctuations one would expect angle independent g⁽¹⁾(τ).

Even though the sample cell was always thoroughly washed before the mea-surements it was unclear whether streptavidin and BSA could contribute the measured signal. Therefore we designed another measurement where we applied the SIAB chemistry (presented in the chapter 4.2.1). By using the sulfo-SIAB chemistry we link thiol-modified DNA directly to the APTES silanized surface over the sulfo-SIAB cross-linker. This means that we should not have any other scatterers in the sample cell than the surface-attached DNA and the sulfo-SIAB monolayer. The measurement was done with a sample cell which is presented in the Fig. 6.8 (a). Here is good seen how the cell is divided into two compartments: the upper compartment was treated with sulfo-SIAB and with thiol-modified DNA and the lower one had only sulfo-SIAB. The measurement were done always in both compartments at angles θ = 35^◦, 45^◦, 55^◦, 65^◦ and 75^◦ with the maximum penetration depth of ∼600 nm, in V-V geometry and a

17Matlab emulation of S. Provencher’s CONTIN program written by I.-G. Marino, (see:

http://www.mathworks.com/matlabcentral/fileexchange/loadFile.do?objectId=6523&objectType=FILE).

single run lasted for 5 seconds. The count rates in the measurements with DNA layer varied between 5 kHz and 20 kHz depending on the adjustment and for the lower compartment we measured low signals being typically below 6 kHz. The results are presented in Fig. 6.7 where in (a) is the averaged and filtered raw data constituting out of 25−108 correlation functions and in Fig. 6.7 (b) is a close up of the scattering signal from the lower compartment. The lower compartment does not show any measurable correlation function. In Fig. 6.7 (c) are the nor-malized correlation functions which were nornor-malized as the data in Fig. 6.6 (b).

The error bars in Figs. 6.7 (a) and (b) are calculated as in the analysis above and present the standard deviation of the averaged correlation functions. The CONTIN results are presented in Fig. 6.7 (d). Furthermore in the Fig. 6.8 (b) and (c) is presented the cell during the measurement. Here is clearly seen that the surface where DNA was incubated is scattering strongly whereas the surface only with sulfo-SIAB surface treatment shows very low scattering.

On the basis of the data shown in the Figs. 6.6, 6.7 and 6.8 it becomes clear that we have measured signal from end-grafted DNA carpet. The data was not further analyzed because of two reasons: (i) as was already mentioned the data is rather noisy and no reliable results could be extracted from it and furthermore (ii) the relaxation time spectrums are difficult to analyze due to several peaks.

The difference between the 2 kbp DNA carpet made on streptavidin surface in comparison to sulfo-SIAB surface could be explained through the fact that probably at the sulfo-SIAB surface DNA has adhered to the surface and we were actually measuring the variations in the static scattered intensity. On contrary at the streptavidin surface DNA is end-grafted and therefore we see a diffuse behavior due to much higher dynamic scattering.