The results from the molecule stretching experiments are presented in Fig. 2.6 where a broad rupture length distribution having average rupture length of roughly 20 µm can be seen. The length of the molecules was measured from the experimental films, error being roughly±5 %8. The presented rupture length distribution was measured with slightly altered silanization protocol (see chapter 4.2.2). This, however, did not seem to have an effect to the result in comparison with silanization presented above (data not shown).
According to Lehner et al. [25] the contour length of a singleλ - DNA, when labelled with YOYO-1 in relation of 1 : 5 (dye molecules to base pairs), is 19.8 µm. This is also the average rupture length in Fig. 2.6. Additionally, as it can be seen from Fig. 2.6, only ∼60 % of the molecules that can be stretched upto the
7Motor: CMA-12CCCL, motion controller: ESP 300, both Newport.
8The error was determined by measuring the length of a single molecule, for 3 different molecules, 10 times and taking the standard deviation which was typically 5 % or less for given molecule length.
buffer + antibleach
objective DNA
glass rod Dx
(a)
Glass rod Motor
Sample cell
(b)
(c)
Figure 2.5: a) The principle of the DNA stretching setup: Functionalized glass rod is brought to contact with glass plate on which DNA molecules are already attached. By moving the rod, the both ends attached molecules are stretched and this is simultaneously observed with the microscope. b) The setup at the microscope. c) A snapshot from experimental data where a single λ – DNA is being stretched. The molecule is roughly 22 µm long. Here it is important to notice that the curvature of the rod is so low that both ends of theλ – DNA are still in focus allowing reliable length determination.
Figure 2.6: The rupture length distribution of stretchedλ- DNA molecules. The measurement was done with the setup presented in Fig. 2.5. The DNA molecules were end-grafted from one end with gold-thiol linkage and from other end with biotin-streptavidin linkage. The average rupture length of the distribution is roughly at 20µm. Data presented with permission from A. Andr´e.
contour length of the YOYO-1 loaded DNA or less. This result means that we have a problem if we want to parallelize the stretching of theλ - DNA molecules.
The rupture of the λ - DNA molecules occurred from both ends, but we did not see any molecules to break along their backbone. Therefore the problem of the surface end-graft should lie either in the silanization, in the crosslinking process of streptavidin to the silanized surfaces, in the biotin-streptavidin linkage, in the gold-thiol linkage or in the ligation of the oligo to the λ - DNA molecule.
The quality of the silanization is very difficult to characterize9 but the proto-cols that we are using are similar or the same as can be found in literature (see for example [26], [18] or [27]). Furthermore, as mentioned above, we did stretch-ing experiments otherwise keepstretch-ing all the same but changstretch-ing only the silanization protocols and we did not observe differences between rupture length distributions.
However, this detail will be further verified in the coming experiments.
The biotin-streptavidin bond is often termed as one of the strongest non-covalent linkages in biology and it has been reported to withstand forces of ∼200 pN [28]. This, however, is not the complete picture. Merkel et al. [29] has mea-sured the biotin-streptavidin bond strength for different loading rates between 1−10000 pN/s. From the measured rupture force histograms for various
load-9We performed some XPS measurements in order to characterize the silanization of the surfaces but these measurements were always unsuccessful due to the carbon dioxide exposure from the air which distorted the results.
ing rates, they determined by gaussian fits to the histograms the most frequent rupture force for a specific loading rate (see Fig. 2.7 (a) and (b)). In order to understand the loading rate dependent strength of the bond we look at this detail closer. For the biotin-streptavidin bond the kinetic on and off rate constantskon and kof f has been reported to have values of 5.13×106 M−1s−1 and 2.8×10−6 s−1, respectively [30]. With these rate constants we can calculate the equilibrium dissociation constantKD =kof f/kon = 5.5×1013M. With the equilibrium disso-ciation constant we can estimate the binding energyEb of the biotin-streptavidin bond as KD = exp(Eb/kBT) where kB is the Boltzmann constant and T is the absolute temperature [31]. This gives us a binding energy of roughly 30kBT and in order to break this bond, a force (f) of roughly f =Eb/x∼200 pN is needed as we mentioned above. Here we have assumed a length of the binding pocketx to be roughly 0.6 nm. However, as the bond is non-covalent, the lifetime of the bond decreases rapidly as force is applied on it due thermal activation. Postu-lated by Bell [32] the dissociation rate of the ligand-receptor complex in solution isτ(0) =τoscexp(Eb/kBT) where τosc is the inverse natural oscillation frequency.
Now when a constant force f acts on the biotin-streptavidin bond the energy landscape is tilted and the activation energy barrier is lowered. The lifetime of the complex is then given by τ(f) = τoscexp((Eb −f xβ)/kBT) where xβ is a distance of the energy barrier from the energy minimum along the direction of applied force (see Fig. 2.7 (c)).
In the Fig. 2.8 is the force-extension behavior of YOYO-1 loaded λ - DNA (measured by Sischka et al. [10]). From this we can estimate our loading raterf since rf is defined as a change of applied force (df) over differential length (dl) multiplied with stretching velocity (v): (rf = (df /dl)v). Now by taking a deriva-tive of the force-extension curve for YOYO-1 loadedλ - DNA and multiplying it with our stretching velocity∼1µm/s. We get an estimate that our loading rates were roughly 5−30 pN/s. Furthermore, asλ- DNA without YOYO-1 shows the transition from B- to S-DNA, it has over very long extension almost unchange-able force, meaning that the loading rate is very low and there is a constant force of 65 pN over the biotin-streptavidin bond. As we now look at the Fig. 2.7 (b), it is obvious that a single biotin-streptavidin bond cannot withstand the forces needed to maintain DNA is the S form.
In order to circumvent the problem of the bond stability we tried also oligos modified with four biotin molecules10. In Fig. 2.9 we show the results of both of the stretching experiments: the red histogram is measured with λ - DNA carrying four biotin molecules whereas the gray histogram is measured with λ -DNA carrying only one biotin molecule (this is the same data as already presented in Fig. 2.6). As is clearly seen from the data in Fig. 2.9 the use of the four biotin in the end of λ - DNA has shifted the average rupture length from 20 µm to 24
10The oligo with four modifications is a custom fabrication by IBA GmbH. The sequence:
(phosphate)-AGG XCG CCG CCC XCX C-biotin. Each X is an T base modified with biotin.
(a) (b)
(c)
Figure 2.7: a) Histograms of rupture forces for biotin-streptavidin linkage as a function of loading rate. Gaussian fits were used to determine the most frequent rupture force from the rupture force histograms. b) The most frequent rupture forces as a function of loading rate for biotin-streptavidin and for biotin-avidin linkages. The star marked data point is measured by Wong et al. [28]. c) The external force adds a mechanical potential that tilts the energy landscape. xβ (see text) is the projection of energy barrier along the direction of the applied force (xβ =hcosθxtSi). All the figures taken from Merkel et al. [29]
Figure 2.8: Force-extension curves for different DNA binding ligands. Our inter-est lies in the curve for YOYO-1 because with this we can estimate the loading rate which we applied to the λ - DNA molecules. Furthermore we can use the force-extension behavior also to give a rough estimate about the forces we applied to the molecules. Figure is taken from Sischka et al. [10].
µm. By comparing this with the force-extension curve for YOYO-1 loaded λ -DNA in Fig. 2.8 we notice that the rupture force has increased from∼10 pN to
∼ 50 pN. Now, if we naively take from the Fig. 2.7 (b) the biotin-streptavidin linkage strength for the lowest loading rate and multiply it by number of biotin molecules, we get roughly 80 pN. However, we estimated the rupture force to be
∼50 pN. How this difference can be explained?
We stretched our molecules always with constant velocity of∼1µm/s whereas Siscka et al. [10] measured their force-extension curve with velocity of 0.1µm/s.
This fact alone makes a crucial difference in force-extension behavior and shows us that we cannot deduce the forces we apply to the molecule out of the Fig.
2.8 (see Siscka et al. [10] Fig. 3B). Further possible error arise from the fact that the equilibration of the YOYO-1 with DNA is reported to last days at room temperature or by incubating at 50◦C for 2 hours [33]. Furthermore, YOYO-1 has two binding modes so that the first binding (bis-intercalation) mode goes up to mixing ratio of 0.125 (dye:DNA base) and at larger mixing ratios the external binding starts to contribute [21]. This means that when incubation times are short the YOYO-1 is mainly attached to the surface of the DNA molecule and the contour length of the YOYO-1 loaded λ - DNA is not well defined as the intercalation between the base-pairs is still going on. This fact was also observed by Bennink et al. [34] who reported elongations of factor 1.1−1.2 when YOYO-1 was incubated only YOYO-15 minutes. We incubated always at room temperature
Figure 2.9: The rupture length distributions of stretchedλ- DNA molecules. The measurement was done with the setup presented in Fig. 2.5. The DNA molecules were end-grafted from one end with gold-thiol linkage and from other end with biotin-streptavidin linkage. The red histogram shows the rupture length distri-bution as the λ - DNA molecules were end-modified with four biotin molecules whereas in the gray histogram λ - DNA molecules had only one biotin in their ends. The gray histogram was already presented in Fig. 2.6. The average rupture length for the multi end-modification was roughly24µm whereas with single end-modification it was only 20µm. Data presented with permission from A. Andr´e.
typically from 15 minutes to several hours, meaning that our YOYO-1 λ - DNA complex was not always fully equilibrated. Sischka et al. [10] do not report their incubation conditions or times making the comparison difficult.
As we analyze the end-grafting chemistry further we found that the crosslinker (glutaraldehyde) which we use to bind streptavidin to the surface is homobifunc-tional meaning that it has the same funchomobifunc-tionality in both of its ends. So upon incubation of glutaraldehyde on the amino silanized surface it can react directly with both of its ends with silane molecules and so the reactivity of the surface against the streptavidin is reduced. This leads to the fact that part of the strep-tavidin molecules are crosslinked over the lysine groups, one or multiple times, to the amino-silane surface and part of the streptavidin molecules are only ph-ysisorbed at the silanized surface. Further problem with glutaraldehyde is that its reaction with amino group containing compound proceeds over several possible routes (see Hermanson [19] page 119). In one of these reaction possibilities the aldehyde end of glutaraldehyde forms Schiff base linkage with amino group and the Schiff base has to be reduced before the bond is chemically stable (covalent).
Here arises the problem that the reduction has to be done after the streptavidin is crosslinked to the surface and the chemicals used for the reduction (for example sodium cyanoborohydride (NaCNBH3)) are also reactive against streptavidin and may harm it [35]. Even if the chemicals were not reactive against streptavidin they would have difficulties finding their way to the bond to be reduced while the streptavidin is sitting on top of this bond [35].
As we already mentioned we have seen also rupture events from the gold-thiol end of the λ - DNA molecules. The gold-thiol bond itself is covalent [19] and should not break upon stretching. Where is the problem then? One potential explanation is that oligos, as they are hybridized to the λ - DNA molecules, are not covalently joined to the rest of the backbone meaning that the T4 ligase has not closed the backbone.
As we look at the data presented in Fig. 2.9 we notice a further interesting detail: we have found few molecules (∼3 %) which we can stretch over 35µm. As was already mentioned the contour length of YOYO-1 labelled λ - DNA is 19.8 µm. This would mean that we have stretched few molecules almost or more than 2 times their contour length. One possible explanation for these long molecules is that upon preparation of the end-modified DNA not every DNA - end is labelled with oligo but some of the molecules join together forming a double λ - DNA.
The remaining questions are: do we really have such double molecules and how big part of the DNA sample contains these double or more λ - DNA molecules also known as concatemers? This question will be discussed further in chapter 3.
As a further interesting detail from the stretching experiments we present Fig. 2.10 where λ - DNA molecule is being stretched between two surfaces11.
11The end-grafting chemistry: Glass rod: Silanized with APTES, treated with glutaralde-hyde, streptavidin and BSA. Bottom plate: silanized with GOPS and reacts directly with
We have highlighted with a red box a part of the stretched molecule where the fluorescent dye is missing. As is clearly seen in the snapshots the dyeless area is within one molecule. However, we do not know whether we are stretching a single λ - DNA or two λ - DNA molecules joined together. This means that we cannot say if we are really stretching the molecule over its contour length. Liu et al. [36] reported that they have seen in their molecular combing12 experiments similar missing fluorescent areas as we present in Fig. 2.10. They explained the missing fluorescence as DNA melting where nicks (backbone breaks) in one of the strands along double-stranded DNA provide a location for a melted single strand to fray back. According their explanation YOYO-1 does not bind on single-stranded DNA and therefore such a melting and back fraying would cause the fluorescent signal to vanish. This, however, is not completely true since for example Auzanneau et al. [37] reported that YOYO-1 binds also on single-stranded DNA. The melting scenario is nevertheless a valid explanation and can be accepted since when the intensity from the intercalated and from the fraying strand is missing from the total intensity, it is reasonable to expect that we cannot detect anymore the single strand of the DNA.
As such an event, where the fluorescence was missing within a stretched molecule, was very rare (less than 1 % of all the stretched molecules), we conclude that we have very little nicks in our end-modified λ - DNA. This is apparently because upon preparation end-modifiedλ- DNA, we add T4 ligase protein to the solution. The ligase protein is used to close covalently the backbone between the DNA molecule and the oligo molecule but it also repairs nicks within the whole DNA molecule [13].
Further analysis of the structure of stretched DNA–YOYO-1 complex is not possible since, as already mentioned above, YOYO-1 binds also to single-stranded DNA and as the strands are staying close together upon stretching we do not see if the DNA is melted without a nick in the backbone. Interesting question here is also why has not anybody reported such an behavior before? The answer is apparently that with tweezers one has limited stretching force of roughly 150 pN [10] whereas we do not have any limit but as already discussed above the exact analysis of the stretching force is not possible due to the missing calibration curve.
Additionally there is a thicker location ”bead” within the molecule and the gap actually arises inside of this ”bead”. We speculate that this ”bead” is apparently a coil of DNA which does not open upon stretching. It could even be that due to the coil of DNA, there is high amount of YOYO-1 concentrated at one location leading to photocleavage of a single DNA strand [38] and thereby provide an explanation for appearance of a nick in the backbone.
amino-modified DNA (see chapter 4.2.2).
12For molecular combing see chapter 5.2.2.
(a) (b) (c) (d) (e)
Figure 2.10: A snapshot series of a single molecule as it is being stretched be-tween two surfaces. The end-grafting chemistry for the glass rod was so that it was first silanized with APTES, incubated with glutaraldehyde and finally with streptavidin and BSA. The bottom plate was silanized with GOPS and it reacts directly with amino-modified DNA (see chapter 4.2.2). Ina)andb)the molecule first elongates as it is being stretched. Inc) the molecule has reached length of
∼22µm and a gap of missing fluorescence is present in the molecule (highlighted with red box). The maximum elongation of the molecule is reached atd) where the molecule is ∼ 28 µm long. Finally after the rupture of streptavidin-biotin bond the molecule relaxes back to its end-grafting site as seen in e). We have added20µm long scale bar in (a) and the size of every snapshot is 6.5µm× 34 µm