LETTERS
NATURE BIOTECHNOLOGY
of DNA in biological organisms compared with their body weight
(~1,000 ppm in Escherichia coli). Finally, we proceeded with 3D
printing of the Stanford Bunny using the same file as stored in the
DNA-containing PCL filament.
Our results show that the data can be perfectly and rapidly
retrieved from the 3D object by consuming a minute quantity of
material using a portable sequencer. We clipped ~10 mg of the
printed PCL from the ear of the bunny, which is 0.3% of the total
material of the bunny that weighed 3.2 g (Fig.
2a). Next, we released
the SPED beads from the embedding PCL using tetrahydrofuran
(THF), extracted the DNA from the SPED beads using buffered
oxide etch (BOE) and purified the library using a standard PCR
cleaning kit. The recovered DNA library weighed 25 pg in 50 µl
volume, corresponding to about 14,000 copies of the encoded
file, including the 5.4× redundancy (Supplementary Note 4). This
entire process took 4 h end-to-end. We then amplified 1 µl of the
recovered DNA, equivalent to 1/50 of the recovered DNA, using
ten PCR cycles and sequenced the library using an iSeq, a portable
Illumina sequencer (Supplementary Note 5). This process took 17 h
and yielded 1,046,118 reads. Finally, we processed the data using
the DNA Fountain decoder and perfectly retrieved the stored .stl
file despite missing 5.9% of the original oligos and being subject to
sequencing errors. This took a few minutes on a standard laptop
(Supplementary Fig. 4).
Encouraged by these results, we conducted multi-round replica
-
tion experiments using the DoT architecture. In the first replication
round, we fused the PCR-amplified DNA from the parent Stanford
Bunny (dubbed ‘P’ following the common notation in genetics) to
a nascent PCL filament using the DoT procedure. Next, we created
three offspring 3D structures (F1) using the PCL filament and the
retrieved .stl file (Fig.
2b). Subsequently, we clipped about 10 mg
from one of the F1 bunnies, extracted the DNA, sequenced the
library with iSeq to retrieve the .stl file and 3D printed the next gen
-
eration. We repeated the same procedure for a total of five genera-
tions (Fig.
2c), where in each generation a new PCL filament was
created by fusing the PCR-amplified DNA molecules of the pre-
vious experiment. Finally, to demonstrate the ability to store the
DNA long term, we sequenced the F4 9 months after its synthesis,
sequenced the information and used it to generate a further product
generation, F5.
We were able to perfectly retrieve the file from all five genera
-
tions of progeny, including the retrieval of information 9 months
after synthesis. In nearly every replication round, we saw an increase
in the fraction of dropout molecules, from a level of 5.9% for the
1001100011
1010100010
1110110100
1110100100
1001001001
Binary stl file
DNA code stl file
DNA
encapsulation
Encode
with DNA fountain
Filament extrusion
Printing of 3D object
DNA extraction
Sequencing and decoding
with DNA fountain
PCR
0110010
1110100
SiO
2
SiO
2
100 ppm particles in PCL 0.2 wt% DNA loading10
5
bunny file units per g PCL
DNA library:
12,000 oligos × 145 nt
Fig. 1 | DoT workflow and proof-of-principle 3D printing of a Stanford Bunny. a, The digital file is encoded into a DNA oligo library using DNA Fountain
encoding. The synthesized library is encapsulated by a sol-gel synthesis method into small glass particles and blended into PCL, which is then extruded
into a standard 3D-printing filament. The object, as defined by the initial digital file, is printed with the PCL filament that contains DNA. The DNA library
can be extracted from any part of the printed object and amplified by PCR. By sequencing the DNA and decoding the DNA Fountain, the original .stl file
can be retrieved to 3D-print new objects. b, Schematic of the hierarchical architecture of the 3D-printed Stanford Bunny, which contains all the instructions
needed to reprint the object. Scale bar, 1cm.
NATURE BIOTECHNOLOGY | VOL 38 | JANUARY 2020 | 39–43 | www.nature.com/naturebiotechnology
40