In situ small-angle X-ray scattering reveals strong condensation of DNA origami during silicification

Silicification of DNA origami structures increases their stability and provides chemical protection. Yet, it is unclear whether the whole DNA framework is embedded or if silica just forms an outer shell and how silicification affects the origami’s internal structure. Employing in situ small-angle X-ray scattering (SAXS), we show that addition of silica precursors induces substantial condensation of the DNA origami at early reaction times by almost 10 %. Subsequently, the overall size of the silicified DNA origami increases again due to increasing silica deposition. We further identify the SAXS Porod invariant as a reliable, model-free parameter for the evaluation of the amount of silica formation at a given time. Contrast matching of the DNA double helix Lorentzian peak reveals silica growth also inside the origami. The less polar silica forming within the origami structure, replacing more than 40 % of the internal hydration water, causes a hydrophobic effect: condensation. DNA origami objects with flat surfaces show a strong tendency towards aggregation during silicification, presumably driven by the same entropic forces causing condensation. Maximally condensed origami displayed thermal stability up to 60 °C. Our studies provide insights into the silicification reaction allowing for the formulation of optimized reaction protocols.

The manuscript by Martina Ober and coworkers investigates silicification process of DNA origami structures by small-angle X-ray scattering. The authors observe interesting effects of DNA origami condensation and contrast inversion during the silicification process. Furthermore, the experimental results indicate that silica deposition occurs both on the outer surface and inner structure of origami constructs. This is detailed and well performed study that provides important insights on the silicification process. The manuscript is clearly written and the data is, in general, well presented. My main issue with this submission to Nature Communications is that while it is often suggested that silica coating of DNA origami structures would help to "unravel their full potential and utilization in real-life applications", in reality, the application of silica coating (which is now well-established) is limited to enhancing stability of origami-based assemblies for structural characterization. Due to this rather limited application, I am not convinced that the manuscript would be of interested to the broad readership of Nature Communications. Publication in more specialized journal seems to be more appropriate in this case.
Other comments -TMAPS induced condensation I am wondering why the measurements were performed only over the span of 8h (Fig  2a, b)? Since DNA origami condensation during silicification process is one of the most interesting aspects of this work, it would be important to characterize at which point the TEOS induced condensation reaches saturation. Related to this, the statement "After eight hours, we obtained a minimal cylinder radius of 73.4 ± 0.4 Å and an interhelical distance of 25.2 ± 210 0.3 Å" is somewhat misleading. The cylinder radius and interhelical distance would probably still decrease with time.
-TMAPS+TEOS induced condensation The data presented on Fig 2d. does not support the following statement "…minimal interhelical distance 23.8 ± 0.2 Å could already be observed after 8 h". There are no measurement points between 0 and 8h, hence, it is impossible to conclude that minimum of interhelical distance is at time point of 8h. -Aggregation.
Aggregation is one of the main problems faced during silicification of DNA origami structures in solution. The authors suggest the origami objects with flat surfaces have strong tendency towards aggregation due to entropic (depletion) forces. This is rather speculative statement; it has been observed in other studies (e.g., ref.13) that DNA origami structures have tendency for tip-to-tip stacking during silicification. As 4LB has 40 helices on the tip vs 24 helices for 24HB, the tip-to-tip stacking might be enhanced. I wonder why the authors did not provide TEM images of 4LB@SiIO2 at the early stages of aggregation, this would provide some insights on the process. Instead, the authors decided to offer rather speculative explanation for the origin of aggregation.
Minor comments -Different folding condition and purification approaches were used for 24HB and 4LB. Why is that? -"All origami discussed so far were cylindrically shaped 24HBs. In order to verify our findings, we also studied cuboid, brick shaped origami during silicification and noted a great tendency towards aggregation,.." was aggregation observed already at TMAPS addition step or only after addition of TEOS ?
We are very grateful for the reviewers' time and their excellent and detailed feedback. We have been able to address all points and we are confident that the manuscript is now ready for publication.

Response to Reviewer 1:
Reviewer: "In this article, Martina F. Ober and his/her co-workers reported a condensation behavior of typical honeycomb structured DNA origami during silicification process by using in situ SAXS experiments and associate data analyzing. Also, they found that silicification takes place not only on the surface, but also inside the nanopores DNA origami, based on tracing the variations of DNA double helix Lorentzian peaks at q ≈ 0.16 1/Å. Overall, this is an interesting work that provides novel insights into the DNA origami templated silicification reaction, and it is important for the understanding of the reactions on the organic-inorganic interfaces.

Response:
We thank the referee for assigning novelty and importance to our report.
Reviewer: However, the major conclusions are too preliminary based on the current data sets. The referee therefore believe that this work could not be published on Nature Communications at its current form, before the following issues have been addressed." Response: In our point-by-point response, we include additional data (S8, S12, S13, S14) as requested by the referee and encouraged by the editor. This data reinforces our claims and hopefully lift any concerns that this work might be preliminary.
Q1> "The authors attempted to draw a general conclusion of the condensation behavior of all DNA origami structures during silicification. But only honeycomb structured DNA origami were tested. Additional experiments on other types of DNA origami structures are necessary."

Response:
We agree that DNA origami come in many shapes and it will be very important to understand if the effects that we observed here are universal. We therefore performed an additional SAXS experiment monitoring the silicification of three-layered blocks (3-LBs) with a square lattice structure (Figure S 10) -the design schematics, the materials and methods, the measurement, and the data analysis are now included in the supporting information, notes S 1, S8, and S13. Similar to the 24HB and 4-LB structures, we find a quickly increasing Porod invariant after the addition of TEOS and a substantial origami condensation during the silicification reaction for the 3-LBs (Figure S 10 bc).

Changes to manuscript:
SI -Page 1, 9, and 13: A new data set of cubic, brick-like structures was added to the supporting information as section S13. The shape parameters by design and the resulting fit parameter, i.e. the condensation effect, was added to the supporting information section S1 Figure S 1and Table S  Main text -Line 291: We edited the following sentence in the main text " […], which we observe not only for origami based on honeycomb lattice arrangements, i.e. the 24HBs and 4-LBs, but also for origami structures based on a square lattice design, i.e. three-layered blocks  as detailed in the supporting information note S8, S13, and S14." Q2.1> "Rigorous SAXS data processing is particularly important for this work. However, over-fitting and fitting plots with high Chi values were observed. E.g., the fitting plots for 4-LBs sample after 10h, especially at q ≈ 0.16 1/Å, were not well agreed with the experiments. These will come to completely different conclusions for the condensation of 4-LBs."

Response:
We fully agree that SAXS data need careful analysis. The analytical fit model, which we use, is quite minimalistic since it essentially reduces the origami to a simple geometric shape, scattering contrast, and regular internal structure. Therefore, the free fit is robust. The problem that the uncertainty is rather large towards longer times (silicification of 4-LBs, 10h and more) is not so much a chi square fitting problem. It is an experimental result: apparently, the 4-LB origami does not take up enough silica to induce contrast inversion reinforcing the Lorentzian peak signal. The pronounced power law behaviour observed in Figure 4a shows very clearly (and without fitting ambiguity) why this might be so. Most of the silica is apparently incorporated into large DNA origami-silica aggregates.

Changes to manuscript:
Main text -Line 274: We edited the following sentence in the main text to clarify our point: " […] but there is no recovery, indicating that uptake of silica is limited." Q2.2> "E.g., Line 272, the authors claimed that, "This aggregation may even obstruct influx of further silica particles into the origami". But aggregation of 4-LBs DNA origami should not alter the intrinsic structural characteristics of single DNA origami structure." Response: Driving particles in assemblies by anti-solvent effects is a widely applied concept in other field of nanoscience (Taylor et al., doi.org/10.1038/s41467-021-22049-8). In this sentence, we wanted to raise awareness that a similar mechanism may be at work here as well for aggregating origami. But we agree with the reviewer that this was not studied systematically so we decided to remove this comment.

Changes to manuscript:
Main text -Line 280: We removed the sentence "This aggregation may even obstruct influx of further silica particles into the origami." Q2.3> "In addition, TMPAS only control experiments were not carried out on this structure. Thus, additional experiment, analysing and re-discussion on the 4-LBs sample group should be provided."

Response:
We performed an additional SAXS experiment studying the effects of TMAPS-only on bare 3-LBs as a representative for a cuboid-shaped origami structure with a square double helix arrangement (Figure S 11). The corresponding data is now included in the supporting information, note S8 and S14. The cuboid height (A) exhibits a dramatic decrease of 30% after an incubation time of 6 h (cf. Figure S 11b). Thus, we observe large TMAPS-induced condensation for various origami structures.

SI -Page 9 and 15: The new data set of TMAPS-only induced condensation of bare 3-LBs was
added to the supporting information as section S14. The resulting fit parameters, i.e. the TMAPS induced condensation effect, was added to the supporting information section S8, Table  S4.

Main text -Line 291: We edited the following sentence in the main text "[…], which we observe not only for origami based on honeycomb lattice arrangements, i.e. the 24HBs and 4-LBs, but also for origami structures based on a square lattice design, i.e. three-layered blocks (3-LBs)
as detailed in the supporting information note S8, S13, and S14." New Figure S 11 SAXS intensities of bare 3-LBs and after the addition of TMAPS together with the best fits of a cuboid model and Lorentz peaks accounting for the inner lattice structure. Data is scaled for clarity. b. Heights of the cuboid 3-LBs extracted from (a) as function of TMAPS incubation time.
Q3> "Porod invariant Q is in inverse proportion to Porod Volume. According to this, the data showed in Figure 1b suggested a reduction of Porod Volume, which, is in contradictory to the slightly increased R value showed in Figure 2c. Please check and explain this contradiction." Response: For the purpose of the Porod analysis, the system here is a two-component mixture (Mantella et al., doi.org/10.1002/ange.202004081). Component one are the silicified DNA origami and component two is the surrounding water matrix. The volume fraction of the silicified DNA is typically called , and we assume the referee refers to this parameter as Porod volume. In this logic the water fraction is (1 − ). The scattering length are and for water and origami, respectively. Then the expression for Q reads According to this equation, a decrease in origami volume could indeed reduce the Porod invariant Q. However, the SAXS data always reveal an increase of the Porod invariant and at later stages a saturation, which we interpret as end of reaction. This indicates that Q is dominated by changes of ( − ) 2 which increases monotonically with uptake of silica. In the supporting information, we originally only emphasized the dominant ( − ) 2 term. However, we see now that we should also include the volume fraction to avoid confusion.

SI -Page 6: "For a two-phase system it is calculated via
Here, the volume fraction of the silicified DNA origami is called and the water fraction is therefore given by (1 − ).
denotes the scattering contrast between the water and the silicificated DNA origami. Thus, provides a measure of the volume fraction and the total scattering contrast, which is in our case the dominant contribution. Thus, monitoring allows to trace the silica growth." Q4> Comparing to previous studies on the sol-gel reaction of DNA origami, the authors reported a more rapid silicification reaction, and reasoned this phenomenal to the tumbling. However, their reagent ratios are different, phosphate group/TMAPS/TEOS, 1:1:15 (Previous work) and 1/5/12.5 (this work). Please provide additional explanations.

Response:
In previous studies on the silicification reaction of DNA origami nanostructures a variety of different phosphate group:TMAPS:TEOS ratios were used (addition of TMAPS in a 1:1 to 1:8 molar ratio (phosphates:TMAPS), addition of TEOS in a 1:10 to 1:104 ratio (phosphates:TEOS)), not just 1:  MgCl2] and the movement of the sample during the reaction (static, moderate shaking) were tested and a significant difference in silicification behavior was observed (Nguyen et al., doi.org/10.1002/anie.201811323). A significant silica growth on the DNA origami nanostructures and thermal stability thereof could only be observed after significantly longer reaction times compared to the silicification reactions with tumbling here in this work (thermal stability already observable after only 4 hours of silicification). In the work by Wang et al. ~ 10 h were required to achieve stable silicified DNA origami lattices.

Changes to manuscript:
Main text -Line 170: We have now added additional reference to the main text. "This is an interesting finding since this time is much shorter than most reaction times reported previously 7,13,14,16 where reactions (employing varying reactant ratios) took up to a week." Q5> "Line 89, silica coating. The silicification reaction protocols for 24HBs and 4-LBs were not identical. Please comment on this."

Response:
We thank the referee for pointing out that we were not clear enough with our description here. During the silicification of cuboid-shaped DNA origami, we observed strong aggregation (milky origami@SiO2 solution) immediately after injection of the reaction mixture (origami + TMAPS + TEOS) into the SAXS sample chamber. We reasoned that this injection via pipette might act as an additional mixing step, which is not present in the original protocol. We therefore decided to inject the second silica precursor TEOS directly in the SAXS sample chamber as a more gentle way to mix the two. In order to clarify this point we changed the text in the manuscript accordingly.

Changes to manuscript:
Main text -Line 98: "Subsequently, TEOS (50 % in methanol) was added 15 min later directly into the SAXS tumbling chamber and incubated directly in the sample chamber to reduce aggregation." Q6> "Line 115 please provide sample-to-detector distance parameter for synchrotron SAXS experiments."

Response:
We now included the sample-to-detector distance (1.7 m) in the "Synchrotron SAXS experiments method" section.

Changes to manuscript:
Main text -Line 122: "Sample-to-detector distance was 1.7 m." Q7> "Line 126, the radiation damage always exists, "reduce the radiation damage" maybe a more cautious claim. Also, please provide the exposure details for each sample."

Response:
We clarify now that the radiation damage is drastically reduced if Mo radiation instead of Cu radiation is used. This was the case for all samples apart from the 24HB@SiO2 heating experiment, which was measured at the synchrotron source. We now included the exposure times for the synchrotron experiments. For in house experiments, X-ray exposure times and silicification time (i.e. time points indicated in the graphs) are largely identical.

Changes to manuscript:
Main text -Line 129: "Mo X-rays induce less radiation dose compared to Cu radiation of the same intensity 27 , allowing for long in situ SAXS experiments with drastically reduced radiation damage to the sample." Main test -Line 120: " […], and an X-ray exposure time of 12 x 10 s." Q8> "Line 129, "larger sample lengths", did the authors refer to "longer optical path distance"?" Response: The reviewer is right, one could also say longer absorption length. In order to clarify, we have changed the text in the manuscript accordingly.

Changes to manuscript:
Main text -Line 132: "Furthermore, Mo radiation allows for larger absorption lengths along the beam (10 mm vs. c.a. 1.5 mm) yielding more practical geometric constrains for SAXS sample cells." Q9> "Line 230, a calculated outer silica shell with thickness of 6.2±0.3 Å was reported. Please consider the molecule size of TMAPS, length of Si-O bond, and draw a detailed molecular scheme for this layer." Response: As per the referee's suggestion, we have now derived an estimation for the expected contour length of silica chains with 1,2,3 and 4 units taking into account all required bond lengths and bond angles. However, silica chains are unlikely fully stretched in a linear fashion, therefore the here calculated values are the absolute upper limit. If branching is taken into consideration, these values fit well with the observed outer shell thickness.