- Research on DNA Data Storage Has Led to the Idea of DNA-Encoded Information in Everyday Objects
- This New Concept is Dubbed “DNA of Things” (DoT)
- Proof-of-Principle for DoT is Demonstrated for 3D Printing
The double-stranded helical structure of DNA was first proposed by James D. Watson & Francis H. C. Crick in a now famous publication in Nature on April 25, 1953, titled Genetic Implications of the Structure of Deoxyribonucleic Acid. Although that report was only one page and had only one figure, which depicted proposed A/T and G/C base-pairing between two antiparallel strands of DNA, its final statement sparked a revolution in the life sciences:
“It has not escaped our notice that the specific pairing we have postulated immediately suggests a possible copying mechanism for the genetic material.”
The now widely recognized central role of DNA sequence information as the “blueprint of life” provides the molecular basis for genetics and personalized medicine, as well as for many tools and methods for the detection, manipulation, editing, synthesis and sequencing of DNA. Continuing advances in the fields of synthesis and sequencing of DNA have led to remarkable progress in synthetic biology, as discussed by uber-famous George M. Church and coauthors in a review titled From Designing the Molecules of Life to Designing Life: Future Applications Derived from Advances in DNA Technologies. For example, there are now biofoundries that provide an integrated infrastructure to enable the rapid design, construction, and testing of genetically reprogrammed organisms for biotechnology research and applications. Examples of notable achievements are provided in an article titled Building a Global Alliance of Biofoundries.
In addition to enabling synthetic biology, low-cost high-throughput synthesis and sequencing of DNA are also being used to explore DNA for information storage and retrieval. According to the Wyss Institute headed by Church, “DNA is at least 1000-fold more dense than the most compact solid-state hard drive, and at least 300-fold more durable than the most stable magnetic tapes”. A recent demonstration of this intriguing approach is provided by Robert N. Grass and coauthors in a 2020 Nature Protocols publication titled Reading and Writing Digital Data in DNA. This article provides the methods and technical details for translating digital information into DNA sequences, physically handling the biomolecules, storing them, and subsequently re-obtaining the information by sequencing the DNA.
In a previous Zone blog, synthetic DNA “barcodes” encapsulated in ammonium-functionalized silica (silicon dioxide, SiO2) nanoparticles were used by Grass and coworkers (Bloch et al.) to tag milk, as well as milk-derived products like yoghurt and cheese, as a means of tracing/verifying the production chain for each item. Grass’ Silica Particles with Encapsulated DNA (SPED) have been shown to mitigate DNA decomposition by heat, oxidants, UV light, etc., while allowing recovery of the DNA by simply using buffered hydrogen fluoride solution (NH4FHF/NH4F) to dissolve the silica. This extraction process is referred to as buffered oxide etch (BOE).
This blog focuses on new research by Grass and coworkers (Koch et al.) that extends SPED labeling to what they call “DNA-of-things” (DoT). As outlined below, DoT involves inserting SPED into manufactured, inanimate objects (“things”) such that the encapsulated DNA encodes manufacturing instructions (“blueprints”) for their replication. This overall strategy for reproducing inanimate objects is intriguingly reminiscent of DNA-encoded reproduction in living cells.
The diagram shown here for DoT workflow, exemplified in the case of 3D printing of an object (see Footnote), is divided into two parts: Make object (steps in blue) and Re-Make object (steps in green). To make the first-generation (1st-gen) of the object, the 3D-printing instructions are converted into a binary stereolithography (stl) file of “ones and zeros” that is then encoded into a corresponding sequence of DNA. The resultant DNA code, which is represented as a pool of chemically synthesized oligonucleotides (oligos), is then encapsulated in silica nanoparticles (SPED) for addition to the 3D-printing material, which in this example is polycaprolactone (PCL).
Once the resultant inventory of the 1st-gen SPED-labeled object becomes nearly depleted, a piece taken from any one of the remaining objects is used to extract the DNA-encoded stl file for PCR amplification prior to sequencing, in order to decode the original binary stl file. The resultant stl file is then used as described above to produce the second-generation (2nd-gen) of the 3D-printed object, which in principle should be an exact copy the 1st-gen object. This workflow can be repeated a number (n) of times to produce an nth-generation object. The magnitude of n will be determined by the extent of accumulated systematic errors, as discussed in the next section, which outlines the proof-of-principle workflow for this overall DoT process reported by Koch et al.
To empirically test data storage using the aforementioned DoT process, Koch et al. created a 3D object that embedded DNA encoding the blueprint for a common computer graphics 3D test model known as the Stanford Bunny, shown here. This was done by compressing the binary stl file of the bunny from 100 kilobytes (kB) to 45 kB, wherein 1 kb is defined as a unit of memory or data equal to 1000 bytes.
Next, these researchers used DNA Fountain—a method that enables robust and efficient information storage architecture—to encode the file in 12,000 DNA oligonucleotides (oligos), which were synthesized electrochemically on a single array of electrodes. With this number of oligos compared to the file size, DNA Fountain encoding yields a redundancy level of 5.2X. This means it can tolerate a dropout of up to 80% of the DNA oligos and still correctly decode the file. The oligos were 145 nucleotides (nt) long, consisting of 104 coding nt and 41 nt for PCR primer hybridization sites.
For 3D printing, shown here, Koch et al. loaded the PCR-amplified oligos into SPED beads and embedded the beads in PCL, [O(CH2)5C(O)]n . This biodegradable thermoplastic polyester offered a low melting temperature and high solubility in various organic solvents, making it an ideal material for blending and 3D printing under mild conditions.
Notably, the PCL filament contained SPED beads at a concentration of 100 mg kg−1 (100 ppm), which did not create any detectable changes to the mechanical properties, weight, or color of the filament. The DNA loading within SPED was 2 mg of DNA per gram of SPED beads, which translates to a DNA concentration of 0.2 mg kg−1 (0.2 ppm) of the PCL filament, well below the concentration of DNA in biological organisms compared to their body weight (~1,000 ppm in Escherichia coli). Koch et al. then 3D printed a 1st-gen Stanford Bunny using the same stl file as stored in the DNA-containing PCL filament.
The researchers demonstrated that the 1st-gen DNA-encoded data could be “perfectly and rapidly retrieved” from a minute quantity of a 1st-gen 3D object by clipping ~10 mg of the printed PCL from the ear of the bunny, which was 0.3% of the total material of the 3.2g bunny. The SPED beads were then released from the PCL using tetrahydrofuran (THF) for subsequent extraction of DNA from SPED beads using BOE. This gave 25 pg of DNA, corresponding to ~14,000 copies of the encoded stl file, including the 5.4X redundancy.
Only 1/50th of recovered DNA was amplified by ten PCR cycles for sequencing using an iSeq Illumina sequencer. This process gave ~1M reads for the DNA Fountain decoder and perfectly retrieved the stored stl file, despite missing ~6% of the original oligos and being subjected to sequencing errors (depicted here as red 0s and 1s).
A video showing all steps in the DoT workflow outlined above can be accessed at this Supplementary Information link in Koch et al.
Multi-Round DoT Replication
Encouraged by these results, Koch et al. conducted multi-round replication experiments using the DoT architecture. In the first replication round, the PCR-amplified DNA from the 1st-gen “parent” Stanford Bunny was incorporated into a nascent, DNA-free PCL filament using the DoT procedure. This PCL filament was then used to 3D print three 2nd-gen “offspring” Stanford Bunnies, which can be seen in Fig. 2b of Koch et al. to be identical progeny.
Subsequently, ~10 mg from one of the 2nd-gen Stanford Bunnies was used to extract the DNA for sequencing with iSeq to retrieve the stl file for 3D printing of the next 3rd-gen Bunny. This same procedure was repeated for a total of five generations. In each one, a new PCL filament was created by fusing the PCR-amplified DNA molecules of the previous experiment. Finally, to demonstrate the ability of long term DNA storage, a 5th-gen Bunny was sequenced 9 months after its printing, and the retrieved DNA stl file was used to generate a further, 6th-gen Bunny.
While all generations of the 3D-printed Stanford Bunny were identical, Fig. 2c in Koch et al. shows a nearly linear increase in the fraction of dropout molecules, from a level of~6% for the original library before inserting the DNA into the PCL filament, to a level of over 20% in the final generation. Readers interested in the detailed analysis of these DNA imperfections can consult the original publication. Koch et al. point out that “despite these myriad imperfections, the DNA Fountain strategy retrieved the file correctly for all generations, showing the robustness of [this] strategy”.
Economy and Versatility of DoT
Cost Analysis: According to Koch et al., the economy of the DoT process is consistent with mass production of goods with memory at negligible cost per-unit. Each replication consumes only 0.3% from each Stanford Bunny and yields sufficient DNA material to create 29 offspring Bunnies. Therefore, even if the number of replications is restricted to five generations, it is theoretically possible to create at least (29/0.003)5 = 8.44 X 1019 bunnies without having to resynthesize the original DNA library.
These results indicate that despite the relatively high upfront cost to synthesize the oligos for the DNA file for the first time ($2,500 in this case), the per-unit cost of synthesis is likely negligible, as mass production will require only one such synthesis event. Koch et al. also conducted an analysis that included the costs of other molecular procedures—excluding sequencing—using current laboratory costs and an industrial scale (see Koch et al. Supplementary Note 6). This analysis indicated that with the industrial scale costs, and manufacturing over 1,000 units of the same item, the cost of a hypothetical 1 MB-containing object (see below) is smaller than the cost of the polymer filament (see Koch et al. Supplementary Fig. 8).
Versatility: Koch et al. envisage that DoT architecture could be compatible with a wide range of embedding materials, their melting temperatures during printing, and the material formation chemistry, which may include radical oxygen species (RoS) or other types of aggressive chemical reactions. This should enable DoT to be embedded in materials manufactured by radical polymerization, such as polystyrene, polyacrylates (see below), and acrylamides.
To demonstrate such versatility, including DNA-encoding scale-up, Koch et al. conducted a DoT architecture experiment in a different context, using a different material, different DNA synthesis methodology, and different material shaping technique. They started by encoding a 2 min video in DNA (1.4 MB), and then used DNA Fountain to encode the file in 300,000 DNA oligos of 104 nt, yielding a redundancy level of 4.2X. These oligos were synthesized by semiconductor-based phosphoramidite chemistry.
The resultant DNA stl file was incorporated into SPED beads that were loaded at a concentration of 100 ppm to methyl methacrylate, which was polymerized using dicumyl peroxide as initiator at room temperature to form polymethyl methacrylate (PMMA, plexiglass), a widely used transparent plastic material. Koch et al. were interested in testing this approach as a means of concealing data, and cast the plexiglass in the shape of a lens for mounting in a frame for eyeglasses. A small portion of the plexiglass eye lens was then used to retrieve the video in perfect condition.
Outlook for DoT
According to Koch et al., the DoT architecture is useful for applications in which blueprint-level information is required to be accessible, either in a personalized item or a mass-manufactured object. For example, in the field of 3D-printed medical or dental implants, as shown here, each structure is unique and customized for the precise anatomical structure of the patient.
As SPED beads are nontoxic, DoT allows for storage of the implant design, as well as other medical background information in it, offering a long-range back-up alternative to traditional electronic medical records, which are usually required to be kept only for 5 to 10 years. Although cost of DNA synthesis may currently be a limiting factor for personalized items, the cost of synthesizing the DNA library becomes marginal for mass-manufactured objects as the number of devices increases.
According to Koch et al., DoT might be applicable to building materials, pharmaceutical products, and electronic components. All of these require product control information, which is currently not available at the point of usage. Having relevant information directly integrated into the product via DoT in an invisible, distributed, and nonremovable manner will consequently allow for better controlled product quality measures. Realizing this vision will require faster DoT retrieval protocols and ubiquitous DNA sequencers, which may be available in the near future based upon rapidly improving, point-of-need nanopore sequencing.
Personally, I am fascinated by all of these results and ideas for DoT applications based on encoding either manufacturing instructions or other valuable information in DNA. I dare say that Watson & Crick did not foresee how their insights into base-paired genetic encoding in DNA, quoted at the beginning of this DNA Day—2020 blog, would eventually lead to the idea of DoT.
As usual, your comments are welcomed.
In a nutshell, the 3D printing process builds a three-dimensional object from a computer-aided design (CAD) model, usually by successively adding material layer-by-layer. This approach, which is also called additive manufacturing, contrasts conventional machining, casting, and forging processes, in which material is removed from a stock item (subtractive manufacturing) or poured into a mold and shaped by means of dies, presses, and hammers. 3D printing enables much faster and cheaper prototyping to test ideas or final manufacturing of either simple or complex objects.
3D printing is now being used to fabricate devices for various types of nucleic acids research, and for printing DNA-containing materials. An example of the former application is described in a publication by Adamski et al. titled 3D Printed Electrophoretic Lab-on-chip for DNA Separation, and an example of the latter application is the subject of a report by Ellington & coworkers titled 3D Printing with Nucleic Acid Adhesives.