'What I cannot create, I do not understand'
This famous quote from Richard Feynman inspired our project. The initial idea of one physicist in our group was:
Let's build a cell!
So of course the biologists of our group tried to explain him, that a cell is far too complex and this is a crazy idea. If a biologist thinks of a cell, he sees an extremely complex structure, but the physicist just sees a simple black circle. So what is a cell, if you break it down to a simpler level?
A compartment separating a reactive center from the outside, combined with controlled means of transport in and out and some modifications on the outside. An intelligent, microscopically small reactor. And our idea of a Smart Nanoreactor was born.
First of all, we needed a compartment. In cells, the compartment is made out of a phospholipid bilayer membrane that forms the plasma membrane. But this being already nature's strategy, we wanted to try a different thing and decided to use block co-polymers to build the nanoreactor wall.
What are polymers?
Polymers consist of structural or repeating units of low molecular weight covalently connected to each other to give high molecular weight compounds. A polymer that consists of two different repeating units is referred to as a co-polymer (Odian, 1981). The covalent linkage of two intrinsically different homo polymers leads to the formation of block–copolymers. Block-copolymers consist of at least two, covalently bound, segments or blocks of different Homo polymers which lead to the formation of di-block or tri-block copolymers. For instance, a tri-block copolymer can have a general form An-Bm-Cp, with A, B, C, being different monomer types constituting the different blocks. The subscripts n, m, and p, stand for the degree of polymerization, i.e. the average number of each monomer present in each respective block (McGrath, 1981).
Why using polymers for our Nanoreactor?
There is growing interest in polymersomes due to their potential application in biotechnology and pharmacology (Blanazs et al., 2009) Currently, a great potential lies in their ability to mimic biological membranes by establishing aqueous three dimensional functional cavities. The stability of these membranes is higher than those of biological lipid-based vesicles. Thus, polymersomes consisting of amphiphilic block copolymers hold promise as alternative vesicles in biotechnological and biomedical applications (van Dongen et al., 2011).
Two important parameters of the membranes are their permeability and their stability. The advantage of synthetic block-copolymers as the building blocks of these nanoreactors is their higher stability over lipids due to the increased length, conformational freedom, and slower dynamics of the underlying polymers.
Other desired characteristics for the Nanoreactor that the polymers can achieve are:
- The shell has a hydrophilic, biocompatible low-protein-binding surface.
- The shell can act as a bio membrane for protein reconstitution.
- The nanovesicle can encapsulate hydrophilic substances.
- The shell material has enormous possibilities for molecular functionalization.
The nanovesicles can thus be employed as specific targeted drug delivery systems (e.g. biotinylated vesicles), stimuli responsive nanocontainers (e.g. pH sensitive) or nanoreactors.
Towards the above goal, two polymers, an ABA tri-block polymer and an AB di-block polymer were investigated.
This ABA tri-block copolymer consists of a middle block of PDMS (poly (dimethyl siloxane) and two PMOXA (poly (2-methyloxazoline)) side chains. These copolymers self-assemble in aqueous solution to form highly stable, closed vesicles with a controlled diameter of typically less than 200 nm.
Tri-block polymersomes were prepared by film rehydration using PBS buffer (pH 7.4) as rehydrating solvent. The polymersomes were observed using cryo-SEM and cryo-TEM, and the particle size was characterized by DLS (Dinamic light scattering).
Cryo-TEM images clearly depict lamellar and large vesicles formation, which is further confirmed by DLS showing an average diameter of ~950nm and a broad particle size distribution. These are not as desired, so these polymersomes were further subjected to an extrusion process using Ultra-filtration equipment.
After extrusion, the samples were observed under cryo-TEM again and their size distribution was measured by DLS. Ultra-filtration process extruded larger vesicles and lamellar structure which is well seen in TEM images. In the DLS measurement, the peak of diameter ranging from 226 to 238 nm is better defined, and the diameter of the bigger particles increase so they can be better separated from the smaller ones. The sample was also inspected using cryo-SEM.
We were also interested in thermo-responsive block copolymer, poly(2-cinnamoylethyl methacrylate)-poly(N-isopropylacryl amide), pH-sensitive and photo-cross-linkable block copolymer. They can function as on/off switches depending on the external pH stimulus due to the pH sensitivity of their cross-linked membrane.
The block copolymers undergo spontaneous self-assembly into the desired polymersome structures by raising the pH of the aqueous solution from 3 to 10. To preserve the vesicular shape of the polymersomes under varying conditions, cross-linking of their membranes is highly desired.
These di-block polymers self assemble into polymersome of uniform size of 100-120nm and later expand in size till 140 nm as the pH is lowered from alkaline condition to acidic state. Thus, they act as pH responsive polymersomes which can tune its release of biological entities based on environmental pH condition. Such property makes them demandable in field of drug therapy specifically in treating cancer.
The formation of the polymersomes was analysed using cryo-TEM and the particle size distribution was measured with DLS. In this case, the distribution is very uniform with a peak at 140 nm.
The pH responsive nature of di-block polymer was investigated by the release profile of Doxorubicin, an anti-cancerogenic agent which showed slow release at pH 7.4 and faster release at pH 5. This is a proof that the polymersomes were effectively cross-linked by UV.
DNA Origami channel
To allow transport inside and outside the smart nanoreactor, a channel is needed. Instead of using a protein channel, we decided to use a DNA origami channel, because it is more robust, and we can easily design it according to the characteristics of the polymer wall.
What is DNA Origami?
In DNA origami a circular ssDNA molecule of several thousand base pairs, called scaffold, is folded with the help of short ssDNA strands, the staple strands. The staple strands bring together two different zones of the scaffold, and by varying their sequences different 2D or 3D shapes can be achieved.
DNA origami has unique properties, such as an addressable surface, which enables selective functionalization with biomolecules and nanomaterials. The origami can also be combined with top-down nanotechnology, such as placement on a fabricated substrate. The technology can also be used in single-molecule imaging, where FRET pairs or fluorophores can be constructed on designed DNA origami structures. Furthermore, DNA origami can be easily combined with other DNA nanodevices such as DNAzymes, DNA beacons or DNA walkers acting on DNA origami. Hence, DNA origami technology has practical potential in various research fields.
The origami channel was designed to be properly integrated into the polymersome wall. For this, the dimensions were restricted to the thickness of the wall, as well as to the length of the scaffold. Furthermore, the dimensions of the porus are determined by the dimensions of the substrate and the enzyme: it has to be big enough so the substrate and products of the reaction can easily go through, but small enough so the enzyme stays encapsulated inside the polymersome. The final dimensions are the following:
- Length: 34 nm
- Diameter: 25 nm
- Pore diameter: 6 nm
Scaffold and staple types
For the design, the caDNAno program was used (see http://cadnano.org/), and the stability of the structure was tested with CanDo (see http://cando-dna-origami.org/).
The scaffold used in this design is 7560 bp long, derived from the M13p18 Escherichia coli virus. Its sequence was taken from the cadnano design program. Bending it into a structure with a 78 helix bundles, the desired length is achieved. The channel contains four types of staples:
- core: 118 staples give stability to the whole structure.
- edge: 39 staples give stability to the edges.
- helpers: 12 staples have a 5' protruding end with 5 adenines, which should help the anchors point to the right directions.
- anchors: 12 staples have a 3' protruding end of 21 nucleotides (7 nm) which will allow the binding of the oligonucleotides that carry the hydrophobic modifications. Between the corresponding complementary sequence to the scaffold and the protrusion, 5 timine nucleotides serve as spacer and hybridize with the 5 adenines present in the helpers.
- no_hang: can be used instead of the anchors to generate a channel without any anchors sticking out of the channel walls. They can serve as a negative control when introducing the hydrophobically modified oligonucleotides.
Their sequences and also the map of their distribution in the scaffold can be donwnloaded from the lab book.
Folding and characterization
The mixture of scaffold and different staples was subjected to a thermal annealing ramp, that allowed the folding. In an initial experiment, different concentrations of Magnesium chloride in the folding buffer were tested, and the resulting structures were analized by atomic force microscopy and electrophoresis to determine the best concentration. The range 10 to 14 mM MgCl2 were the best, so we chose the 12 mM concentration for further experiments.
In the AFM images the height and width were measured, obtaining the approximate values of 7 and 60 nm respectively. The height appears to be much lower than the expected, due to the high absorption of the channel to the substrate used for imaging, mica. Mica is negatively charged, and so the Magnesium chloride in the samples interacts with the surface creating a positively charged substrate. Thus, the DNA origami channel can interact tightly with the surface, modifying its structure. The width appears much bigger, which can be due to the tip-sample convolution.
The 3D structure was then corroborated using TEM and cryo-TEM.
For the TEM images, the diameter and length of 20 different structures was measured. In this case, the dimensions of the channel and also its shape correspond accurately to the design.
For the channel to be incorporated inside the polymersome membrane spontaneously, a hydrophobic modification was introduced to a short oligonucleotide that hybridizes with the anchors. Two different types of hydrophobic modifications were performed.
5' Palmitate and arachidic acid modification
An amino group was introduced to the 5' end of the oligonucleotides, that afterwards reacts with the carboxylic group of the molecule containing the hydrophobic fatty acids palmitate and arachidate. The reaction used is an N-acylation, which requires some activations. More details of the reaction and reagents used can be found in the Lab book.
The modified oligos were purified with HPLC and the purity of the product was analized by mass spectrometry.
3' Cholesterol modification
Oligonucleotides for 3' cholesterol modification were synthesized on a solid support already carrying the cholesterol molecules. In this case the sequence was shorter for the modification to be slightly separated from the walls of the channel. The modified oligos were purified with HPLC and the purity of the product was analized with mass spectrometry.
Incorporation of the hydrophobically-modified oligonucletides into the origami channel
The success of the incorporation of the hydrophobically modified oligos into the DNA origami channel was analyzed by electrophoresis.The DNA origami channel was folded with the presence of no_hang staples, so a structure with no anchors was produced. This channel without anchors was put into contact with the oligonucleotide complementary to the anchors without modification, and also with the oligonucleotide with the two different types of hydrophobic modifications. This served as a negative control as no difference in the electrophoretic mobility should be observed.
In parallel, the DNA origami channel with the anchor staples was produced in the present of the oligonucleotide complementary to them without the hydrophobic modifications, and the same oligonucleotide with the two different types of modifications
In the electrophoresis gel a shift can be observed between the electrophoresis mobility of the channels that don't ocntain the anchors and the ones that contain them. This is because the shape of the structure is affected by these sticking out DNA strands in a way that they cannot run as fast in the gel. Moreover, when introducing the oligonucleotides that hybridize with the anchors 5' modified with palmitic and arachidic acid, a further shift is observed, indicating that this modification is correctly incorporated into the channel. As for the 3' cholesterol modification, no shift is observed, but this can be due to the small volume of the cholesterol molecules in respect to the channel and their situation closer to the channel walls. In both modifications, though, a remaining of the sample inside the wells can indicate a certain level of formation of big complexes not able to enter the gel due to aggregation of the channel through their hydrophobic regions, specially in the 5' palmitic and arachidic acid modified channel. This is a good inticative that the hydrophobic modifications were properly incorporated onto the DNA origami channel.
The aim of our project is to create a nanoreactor. Inspired by how cell metabolism is organized in specific cellular compartments, we decided to design a system which is capable of hosting enzymatic reactions. Inner compartmentalization allows fragile processes to remain protected against undesired influences, such as proteolytic or microbial degradation and also one could expect enhanced reaction probabilities and efficiencies due to spatial confinement or immobilization of enzymes.
Choice of reaction
There are several requirements for the enzyme which can be entrapped. The enzymes should be large enough so that they are not able to escape, especially if there are channels or pores formed in the system. Another requirement is that the reaction is easy to detect. Examples include enzymes which are able to produce fluorescent or chromogenic products, such as myoglobin, Candida Antarctica lipase B, glucose oxides, horse radish peroxidase, urease and α-chymotrypsin.
The protein of our choice was collagenase. Collagenases are endopeptidases that degrade the helical regions in native collagen preferentially at the Y-Gly bond in the sequence Pro-Y-Gly-Pro, where Y is most frequently a neutral amino acid. This protein is very abundant in many species and it is easy to obtain from the bacteria Clostridium hystoliticum. The isolated enzyme is composed of 7 different proteases ranging in molecular weight from 68-130 kDa, thus the protein is big enough and there is no possibility to pass through the channel. Furthermore, the reaction can be detected easily with spectrophotometric assay.
The protein was conjugated to amine-reactive fluorescent dye in order to prove its entrapment inside the polymersomes. TAMRA NHS was used to perform the fluorescent labeling. TAMRA (Tetramethylrhodamine) belongs to the group of the long-wavelength rhodamines with excitation maximum at 555nm and emission maximum at 580 nm.
The degree of labelling with TAMRA dye was estimated to be 18 dye molecules per enzyme.
Enzyme reaction assay
The enzymatic assay for determining the rate of the reaction was carried out using the peptide FALGPA, standing for N-(3-[2-furyl]acryloyl)-Leu-Gly-Pro-Ala. There is sufficient structural similarity between the 2-furanacryoyl group and Pro for FALGPA to be an excellent substrate for collagenase.
The molecular weight of the substrate is 476 Da. The maximum molecular weight of the molecules that can go through the polymer walls being 400 Da, the substrate will not be able to penetrate the polymer walls, but only to enter the nanoreactor through the channel. Furthermore, the substrate being uncharged, the negative charge of the DNA won't interfere with its transport.
This peptide has absorbance at 345 nm and the reaction is monitored through continuous spectrophotometric measurement at this wavelength. The assay is made possible by the blue shift in the near-ultraviolet absorption band of the furanacryloyl peptide when the peptide bond between the first and second residues is hydrolyzed, which means that after cleavage of the peptide, the absorbance at 345nm begins to decrease. The enzyme activity before and after conjugation to the dye is characterized through this assay.
The conjugation to the fluorescent dye leads to a slight decrease in the activity of the enzyme. The FALGPA units are calculated, taking into account the slope of the linear fit. The unmodified enzyme is estimated to have 3 FALGPA units/ml and the conjugated one – 2.6 FALGPA units/ml.
Since the polymersomes can be source of light scattering events, UPLC measurement of the reaction products was applied for better accuracy of our experimental results.
Incorporation of the channel
The DNA origami channel was incorporated during the polymersome preparation and was characterized using TEM and Cryo-TEM.
For the tri-block copolymer, the polymerosome formation was not affected by the addition of the DNA origami channel, and the DLS analysis confirmed that the polymersome size distribution was the same. In some of the polymersomes, particles across the membrane were observed. This indicates that the origami channels were successfully incorporated.
To further confirm the incorporation of the channel in the wall, gold nanoparticles can be attached to the DNA channel. This will make them more distinguishable in the TEM and cryo-TEM analysis. This is planned as a follow-up experiment.
The functionality of the channel was tested through a dye release assay. As a further confirmation of the incorporation of the channel in the membrane, and its functionality, we performed a dye release assay.
For the di-block copolymer, the polymersome formation seemed to be affected by the presence of the DNA origami channel. The polymersome membrane surface became rough and also its stability decreased.The polymersomes had a smaller diameter than previously observed and were present in a smaller amount.
Dye release assay
In order to confirm the functionality of the origami channels and their successful integration into the polymersome walls, a dye release experiment was performed. The fluorescein dye was encapsulated into the polymersomes in presence and absence of the channels. Fluorescein molecular weight is 332.29 Da and it can pass through the polymer walls, although with some difficulties. Thus, the incorporation of the channel should facilitate the exit of the dye. All the samples were placed inside dialysis bags and the fluorescein concentration outside the polymersomes was measured by UV/vis spectroscopy.
As a positive control, the release of fluorescein from the dialysis bag without polymersomes was also performed.
Tri-block polymersomes show around 80% of release within 24h whereas trib-lock polymersomes with DNA channels exhibit complete release of dye. This is a strong hint of the successful incorporation of DNA channel within the tri-block polymersome walls, and that those are functional and allow the fluorescein to go through.
Di-block polymersomes with and without origami channels have similar release profile. The release of dye from the di-block polymersome was less than 50% which proves association of dye with polymer. It is in agreement with confocal LSM images presenting strong adhesion of the dye molecules with di-block polymer which may inhibit polymersome formation. In this case, thus, the dye-release assay cannot be used to prove the incorporation of the channels due to the interaction of the dye with the polymersome walls.
After 24h, the samples were also imaged by C-LSM (Confocal Laser Scanning Microscopy).
Tri-block Polymersomes without channels clearly trapped the dye molecules which could be observed in C-LSM images, whereas the polymersome with DNA channels released most of the dye and residual polymers containing dye could only be imaged with High Laser power and increased gain.
Di-block polymersomes were highly destabilized in presence of dye. Only a few polymersomes encapsulated dye molecules similar to tri-block polymersome. Most polymersome were in disrupted state and strongly adhering fluorescence dye molecules. This result is in agreement with cryo-TEM images were we observed the similar structure of polymersome.
Incorporation of the enzyme
The collagenase conjugated with the TAMRA fluorescent dye was encapsulated in both polymerosomes. Dialysis was performed to eliminate the non-encapsulated enzyme. The encapsulation efficiency was calculated as the concentration after dialysis divided by the concentration before dialysis. For the Tri-block polymersomes, it was of around 70%, whereas for Di-block polymersomes, it resulted to be of around 30%.
The samples were then imaged by C-LSM. Particles that correspond to the polymersomes with the encapsulated enzyme could be observed in both Tri-block and Di-block polymers.
In order to prove that after dialysis all the free enzyme is removed, the polymerosmes were incubated with substrate solution and UPLC measurements were taken at different intervals of time. Even after 16h of incubation, no products were detected, meaning that no free enzyme is present in the solution and that the enzyme is successfully encapsulated inside the polymersomes.
Incorporation of enzyme and channel
As we demonstrated that both DNA channel incorporation and enzyme encapsulation was successful. In focus to our smart nano-container with these dual functions, the assembly of DNA channels and encapsulation of active enzyme at the same time is to be demonstrated. We choose enzymatic analysis to prove chemical transformation by FALGPA substrates which enter the polymersome through integrated DNA channel and get converted into products which can be detected with high sensitivity using UPLC (Ultra Performance Liquid Chromatography).
In presence of enzyme, the DNA channel incorporation into membrane of polymersome was found to be successful shown using cryo-TEM image. We hypotheses the integration of DNA channels into the polymersome membrane from image contrast modulations around the polymersome membrane which are definitely not present in absence of DNA channel or Enzyme. In addition, the channels were also found to be perpendicular attached to the polymersome surface which may or may not be associated with polymersomes. Further, confirmatory results can be obtained by three dimensional cryo-TEM image collection and reconstruction.
In order to obtain further evidence for DNA channel incorporation into polymersomes, negative stained sample were inspected under TEM. Although one should consider that the polymersomes get atleast partly collapsed during this preparation the staining of DNA channels with and without tri-block polymer shows significant morphology differences. In absence of polymer, the DNA channels are clearly visible as cylindrical motif with sharp defined contours. While in presence of polymer, the DNA channels are embedded in a diffuse matrix addressed as polymer shell which is attached to the channel.
In order to confirm our hypothesis, enzymatic analysis was performed and analyzed using UPLC.
As we observed in DNA channel incorporation, the di-block polymersome represented the similar morphology with highly distorted membrane almost present in fragments rather intact polymersomes.
Internal Reaction assay
The enzymatic assay was performed to prove that our system works, and so the channel allows the transport of the substrate in and out of the polymersomes, and the encapsulated enzyme is still active and the reaction takes place.
The reaction assay was performed after dialysis of the samples, so that no free enzyme is present in the solution. For the samples with DNA channel, two type of reaction were set – one with bigger concentration and another with smaller concentration of the polymersomes.
From the chromatograms can be seen that there is conversion of substrate into product after 3h incubation. However, the reaction rate is slower because even after 48h the uncleaved substrate is still present and it is more than the obtained product.
For detecting the reaction products UPLC measurements were conducted. There is a correlation between the data from the cryo-TEM and from the reaction assay. Only for the samples which showed better incorporation of the channel, reaction products could be detected. The samples with di-block polymersomes, showed no reaction at all, even with higher concentration of polymersomes. For the tri-block polymersomes we could observe presence of reaction products with both concentrations. Nevertheless, for this type of polymersomes reaction could be seen in the sample without channel, which perhaps is due to not sufficient time of dialysis. For this reason, the results are not conclusive about the efficient incorporation of the channel and further optimization of concentration and time for dialysis needs to be done.
Drug delivery system
In recent years DNA origami reached incredible popularity in nanotechnology due to its vast design possibilities and its broad potential applications. While having numerous advantages it is still exceptionally expensive to produce in high amounts for mass applications, whereas inorganic polymers, another range of materials available for biomolecular design can compensate for this disadvantage. Nowadays they can be found in almost every aspect of our lives, in medicine and technology through their wide range of different functionalities and their moderate price. In our project we aim to combine those two artificial materials into one device: functional and responsive polymers as carrier material with highly flexible designable DNA origami for more complex access points to gain the best of both of them.
Our main application presented here in our video we describe using our Smart Nanoreactor as a new form of drug delivery system, which takes great advantages of the polymersome´s role as carrier for multiple different other functional components. In chemotherapy a big range of treatment methods use drugs which are damaging both cancerous as well as healthy tissue, putting the patient in a high stress during the treatment. But with out Smart Nanoreactor we offer a solution for this problem. As a first step the Nanoreactor is injected into the patient and through antibodies immobilized on the outside of the sphere, the device is targeting specific pathogenic tissue like cancer and is localized only in this area. In the second step an inactive form of a pro-drug gets systematically introduced into the patient, so it is localized in healthy as well as pathogenic tissue. Now the pro-drug can enter the Nanoreactor through the origami channels and the encapsulated enzyme can convert the inactive pro-drug to its active state in which it is damaging both healthy and pathogenic tissue. After exiting the Nanoreactor again the active drug can interact with its surroundings, due to the location of the Nanoreactor the closest cells are the intended cancerous tissue. After the targeted cells are treated the immobilized antibodies loose their binding partner, the reactor dis-attaches and is filtered out of the body through the kidneys.
Another powerful feature of our device is the use of a pH-responsive polymer, which enables us to add even more functionality to our Nanoreactor. The described Di-block copolymer has tight and stable walls in basic pH but gets leaky in acidic milieu. We could use this to create another route of transport for molecules that are too big to fit through the origami channel. In high pH the container circles through its environment and takes up small molecules from the outside, an enclosed enzyme crosslinks them together until they are too big to exit again through the channel. After the container reaches surroundings with lower pH, its walls become more porous and allow the exit of its cargo through pores in the polymersome. A great advantage is the reversibility of this process, so when the vesicle is empty, it can be brought back to basic pH, the walls are sealed tight again and the reactor can take up more substrate through the DNA channel.
This feature could not only open a variety of applications in medicine where our device can react to physiological changes in pH but also in technology and environmental engineering for example. Synthesizing processes can be automated and our Nanoreactor can respond to different stages of a pH changing reaction or it could be used to clean freshwater supplies via independent sampling of contaminated wells.
The possibilities of our device for technological applications do not end here but observing nature and discovering the formation of polymersomes inside bigger polymersomes during Cryo-TEM analysis another great range of technical applications came to our minds. If we extend this process to creating complete Smart Nanoreactors encapsulated in other Nanoreactors we can create whole Nanofactories in the future, similar to the different lipid vesicle compartments in living cells like vacuoles.
Combining different reactors with various different DNA origami channels, polymersomes and enzymes we are able to design a system, which can independently catalyze complex chains of reactions. Immobilized targeting proteins on the outside of polymersomes and their sizes act for sorting of the reactors, while the different channels determine the order of entry that the substrates have to use to complete the full chain of a reaction. Besides easy separation of product molecules form enzymes, our device has the advantage of formation of big structures by different enzymes in a complex order, one can even create repeats and loops by incorporating a range or Nanoreactors within each other, all of it independently.
This was just a small trip through the wide spectrum of applications for our Smart Nanoreactor, but there is far more possible in the future. And all of this just with a plastic ball.
- A. Blanazs, S. P. Armes, A. J. Ryan, Macromol. Rapid Commun. 2009, 30, 267; b) C. LoPresti, H. Lomas, M. Massignani, T. Smart, G. Battaglia, J. Mater. Chem. 2009, 19, 3576.
- S. F. M. van Dongen, W. P. R. Verdurmen, R. J. R.W. Peters, R. J. M. Nolte, R. Brock, J. C. M. van Hest, Angew. Chem. 2010, 122, 7371; Angew. Chem. Int. Ed. 2010, 49, 7213; b) K. Renggli, P. Baumann, K. Langowska, O. Onaca, N. Bruns, W. Meier, Adv. Funct. Mater. 2011, 21, 1241; c) F. H. Meng, Z. Y. Zhong, J. Phys. Chem. Lett. 2011, 2, 1533.
- G. Odian, Principles of Polymerization. 2nd Ed, 1981.
- J. E. McGrath, Journal of Chemical Education 1981, 58, 914
- P. Rothemund, Folding DNA to create nanoscale shapes and patterns. Nature 2006, 440, 297-302.