No evidence is found for the specific detection of clusters of

In numerous force maps recorded in MR-AFM experiments clusters were identified. This holds for measurements with conventional IgG antibodies raised against syntaxin-1 coupled to the cantilever (Figure 4.8) and with coupled nanobodies against syntaxin-1 (section 4.1.5.2 and Figure 4.15 therein and section 4.1.7). When free IgG antibodies against syntaxin-1 were added to the measuring buffer a significant decrease in the frequency of events was observed as compared with a measurement on exactly the same membrane sheet before the

5.1 Heterogeneity and Clustering in PC12 Membrane Sheets

addition. This is the strongest evidence obtained that specific interactions might have been detected. However, still many events were present after the addition of the antibody. Such a finding is reasonable, especially at long dwell times since the local concentration of the nanobody on the cantilever is high and a substitution of the free antibody is favoured entropically. An even more pronounced non-effectiveness of the competition of antibodies when measured by conventional force mapping was described by Creasey et al. when they performed MR-AFM experiments on human lenses.^[101] To solve this problem they suggested to use TREC mode imaging instead.

For antibody-antigen interactions most probable rupture forces of about 40-60 pN are reported.[28,99,101] In MR-AFM measurements from 15 maps of the present thesis recorded with anti-Syx-NBs coupled to gold coated cantilevers (Figure 4.14) this value is about 30 pN.

For experiments with anti-Syx-NBs coupled to cantilevers which were maleimide functionalised by the manufacturer the most probable maximum interaction force is about 65 pN, irrespective of the type of membrane sheet used for the measurement (Supplementary Figure 5). The latter measurements were performed with a five times increased retraction speed, which might explain the larger forces. When compared with the values reported in literature, it has to be kept in mind that the cantilever retraction speed influences the unbinding force. Unfortunately, except for the study of Creasey et al.^[101] who recorded their force curves with 400 nm∙s⁻¹, no cantilever retraction speeds or loading rates are reported in the cited literature. Under these circumstances, the most probable forces observed might be compatible with the detection of specific interactions. However, when regarding the direct comparison between measurements on wild type and knockdown membrane sheets (Figure 4.24) the similarity in the force distributions does not point to a detection of specific interactions in the case of wild type sheets.

The median rupture distance obtained from MR-AFM measurements on PC12-WT-2 membrane sheets with cantilevers which were maleimated by the manufacturer (33 nm) is much closer to the contour length of the used PEG spacer (M = 3.4 kg∙mol^–1) of almost 30 nm^[30] than the median rupture distance observed on knockdown membrane sheets (83 nm) (Figure 4.26). As to be expected, the same tendency is observed for the contour lengths (Table 4.4) even if the absolute numbers are almost one order of magnitude larger than expected. A slightly longer rupture distance than the contour length of the PEG chain is reasonable since the size of the nanobody and that of the target protein contribute as well.

The shorter rupture distances detected on wild type membrane sheets may indicate that the structures which induced the unspecific events depend on syntaxin-1. Alternatively, it might be explained by specific interactions with the PC12-WT-2 membrane sheets whereas for PC12-Syx-KD sheets, where presumably no specific interactions can occur, unspecific

interactions with filamentous or other elongated structures take place instead (see section 5.1.3), therefore giving rise to a larger rupture distance. However, such an effective competition for unspecific events is unlikely and can almost be excluded by other findings obtained in the present study (vide infra). Furthermore, this tendency is not observed when cantilevers which were just aminated by the manufacturer were used instead (Table 4.2).

This might, however, be explained by a non-successful reaction between the aminated surface and the N-hydroxysuccinimide in the latter case. Alternatively, the genetic interference of the knockdown might have resulted in looser and more elongated structures which cause the unspecific events. Especially due to the loss of syntaxin-1, which can bind other proteins^[73] and tether vesicles^[5], a degeneration of the structures on the inner leaflet which leads to a looser organisation of the entire structure is reasonable. This hypothesis could be probed by performing MR-AFM measurements on membrane sheets derived from control PC12 cells which have been transfected with an empty vector corresponding to the one used for the knockdown but which does not induce a knockdown. Additionally, MR-AFM experiments with nanobodies raised against mCherry might also be performed on PC12-WT-2 and PC12-Syx-KD membrane sheets.

It has to be noted that the concentration of nanobodies used for the competition experiments was almost two orders of magnitudes smaller than suggested by Hinterdorfer.^[154] However, in their first study Hinterdorfer et al. described a successful competition by using a concentration of antibodies which was only a factor of 1.5 larger than the concentration of nanobodies used for one competition experiment in the present study.^[93] When in an initial experiment nanobodies, which had not been subject to treatment with TCEP, were added to the measuring buffer in a final concentration of 1 µ^M a large frequency of events with multiple peaks in the force curves was observed. Thus, the concentration was decreased and TCEP treated nanobodies were used for further experiments. Furthermore, this indicates that added nanobodies might be capable of inducing further interactions between the cantilever and the substrate on their own, as described in section 4.1.6.1.

A further source of cluster detection, even when only to a small extent, might be caused by the choice of the threshold value in the Ripley analysis: Due to the adaption of a threshold value of the mean plus three standard deviations of the Lj(r)–r values from 100 homogeneous random distributions even for a homogeneous distribution of events six pixels with Lj(r)–r values above the threshold are expected to be detected in a 64×64 grid.

It cannot completely be excluded that a cluster of remaining syntaxin-1 molecules in PC12-Syx-KD membrane sheets was detected, e.g. in Figure 4.25 d, since a minor amount of syntaxin-1 was still detected in PC12-Syx-KD sheets. However, the supposed distribution of syntaxin-1 in PC12 membrane sheets measured by super resolution microscopic methods

5.1 Heterogeneity and Clustering in PC12 Membrane Sheets

such as STED (see Figure 4.4) or direct STORM^[90] revealed zones in membrane sheets depleted of syntaxin-1. Thus, there are also in membrane sheets derived from wild type cells regions with a sparse number of syntaxin-1 which might be comparable to that of knockdown cells. It is reasonable that some measurements on wild type membrane sheets were performed in such a region. This renders the above mentioned effect very unlikely.

Furthermore, the frequency of events is not reduced in PC12-Syx-KD membrane sheets, in contrast to the expectation.

Strong evidence against the concerns that clusters detected by control experiments with antibody or nanobody competition or knockdown cells might be caused by remaining specific interactions can be gained from experiments performed with nanobodies raised against mCherry coupled to the cantilever. Here, in two independent experiments not only interaction events were observed but also identified to be segregated in clusters.

For most experiments it was attempted to compare frequencies of events between control measurements and those suggested to provide specific interactions. First, it has to be noted that for a few force curves the presence of an interaction event could not be evaluated and therefore the overall amount of events is slightly underestimated, because no event was assigned to the corresponding location. In most cases the determined frequencies were not significantly different or resulted only in slight differences and sometimes even indicated a larger frequency of events on syntaxin-1 knockdown membrane sheets as shown in section 4.1.7. The main contributor to the variation of fe is probably the cantilever itself (Figure 4.23). This renders the obtained frequency of events kind of speculative, therefore direct comparisons can only be used for estimations, unless a larger number of membrane sheets or GUV-MPs and cantilevers were used for their determination. Nevertheless, the variations among different membrane sheets when using the same cantilever (Figure 4.23) indicate that membrane sheets of different quality are present which induces varying amounts of presumably unspecific events, which might be related to the structures residing on the membrane sheets.

Finally, by the threshold analyses described in sections 4.1.8.2 and 4.1.8.3 no differences between maps of events obtained from PC12-WT-2 and PC12-Syx-KD cells could be identified, either. For the only map of wild type membrane sheets yielding a cluster which was not identified without the threshold a large force threshold of Flow = 468 pN or a large lower threshold of the contour length of Lc,low = 2.94 µm resulted in the maximum value of L(r)–r. Taking the expected maximum interaction force into account (vide supra) this value is unreasonably large. Thus, so far no additional clustering could be inferred in wild type membrane sheets which is not found in knockdown sheets similarly. However, the results of these analyses can only be regarded as preliminary, more intense analysis would be

necessary to certainly exclude a discrimination between membrane sheets derived from wild type and those from knockdown cells. Especially the large difference of the rupture distances between wild type and knockdown membrane sheets identified in Figure 4.26 b might be the most promising object of further analyses.

Taken together, only sparse evidence for a detection of specific events is found which are almost exclusively negated by other findings. Thus, it cannot be stated on the organisation of syntaxin-1 in PC12 membrane sheets.

5.1.3 Reasons for large ratio between unspecific and specific events remain