WO2003025123A2

WO2003025123A2 - Dna: a medium for long-term information storage specification

Info

Publication number: WO2003025123A2
Application number: PCT/US2002/027606
Authority: WO
Inventors: Carter Bancroft; Catherine Clelland
Original assignee: Mount Sinai School Of Medecine
Priority date: 2001-08-28
Filing date: 2002-08-28
Publication date: 2003-03-27

Abstract

The present invention relates to a technique for using nucleic acid molecules as a medium for long-term storage and retrieval of information.

Description

DNA: A MEDIUM FOR LONG-TERM INFORMATION STORAGE

SPECIFICATION

INTRODUCTION The present invention relates to a technique for using nucleic acid molecules as a medium for long-term storage and retrieval of information.

BACKGROUND OF THE INVENTION

In this digital age, the technology employed for information storage is undergoing rapid advances. Data currently being stored in magnetic or optical media will in all likelihood become unrecoverable within a century or less, through the combined effects of hardware and software obsolescence, and decay of the storage medium. New approaches are required that will permit retrieval of information stored for centuries or even millennia.

SUMMARY OF THE INVENTION The present invention relates to a technique for using nucleic acid molecules as a medium for long-term storage and retrieval of information

BRIEF DESCRIPTION OF THE DRAWINGS

Figure 1. Structures of DNA Molecules Employed for Information

Storage and Readout. The single Polyprimer Key contains a series of sequencing primer sequences ("Seq Primer #") flanked by common forward ("F Primer") and reverse ("R Primer") PCR primer sequences, each separated by a small common spacer. Each Information DNA is also flanked by the common PCR primer sequences, and contains two unique elements, separated by the common spacer: a Seq

Primer and a numbered information segment ("Information #") (not drawn to scale). Figure 2. Experimental Prototype. Readout of information DNAs, including sequencing plus decoding. DETAILED DESCRIPTION OF THE INVENTION

DNA possesses three properties that recommend it as a vehicle for long-term information storage. First, DNA has stood the informational "test of time" during the billions of years and multiple generations since the early days of life on Earth. Non-replicating DNA molecules have proven to be quite robust. Although DNA stored under non-ideal conditions, e.g., in ancient archaeological deposits, is subject to hydrolytic and oxidative damage (Moss, M. et al., 1996, Nucl. Acids. Res. 24:1304) mitochondrial DNA extracted and amplified from 7000-year-old human remains yielded an accurate DNA sequence (Paabo, JA et al., 1988, Nucl. Acids Res. 16: 1775). Storage of DNA under more ideal conditions can result in extremely long stability, as evidenced by the reported recovery of viable bacteria (and thus of their genomic DNA) from 250-million-year-old salt crystals (Vreeland, RH et al., 2000, Nature 407:897). Second, since DNA is genetic material, methods for both storage and reading of DNA-encoded information should remain a central core of technological civilizations, and undergo continual improvements in both efficiency and miniaturization. Third, theuse of DNA as a storage medium permits each segment of information to be stored in an enormous number of identical molecules. This extensive informational redundancy strongly mitigates effects of any losses due to stochastic decay. Studies of ancient human remains provide information on long-term rates of DNA decay and/or modification, and thus a highly conservative estimate of minimum DNA amounts required for prolonged storage under more ideal conditions. About 0.1% of the DNA extracted from ancient, decomposed tissue is unmodified (S. Paabo, R.G. Higuchi, AC. Wilson, J. Biol. Chem. 264, 9709 (1989). However, as little as 100-300 fg of this unmodified DNA can serve as a PCR template and generate accurate DNA sequence. In the example described herein, information is stored in 20 ng (20,000 pg) of identical DNA molecules -250 bp in size, far above the above the 100 pg range. Moreover, use of this large number of molecules (c. 80 billion) as PCR templates in the readout procedure greatly suppresses effects of any base modifications during prolonged storage that yield sequence changes in individual DNA molecules (O. Handt, M. Krings, R.H. Ward, S. Paabo, 1996, Am. J. Hum. Genet. 59, 368. An optimal procedure for long-term storage of information in DNA is designed so that data retrieval requires minimal prior knowledge beyond a familiarity with molecular biological techniques. The method of the present invention utilizes two standard procedures for recovery of stored information: polymerase chain reaction (PCR) and DNA sequence analysis. The method uses two classes of nucleic acid molecules as depicted in Figure 1.: (i) "Information DNAs" (iDNAs) containing the stored information, and (ii) a single "Polyprimer Key" (PPK) that is the key to retrieving the information stored in the iDNAs. Each iDNA contains the following sequence elements: common flanking forward (F) and reverse (R) PCR amplification primers (~10-20 bases long), a unique sequencing primer (Seq Primer) (comparable in size to the F and R primers), a small common spacer (~3-4 bases long) serving as a cue to indicate the start of the stored information, and a unique information segment. The information to be stored is encoded successively in these information segments, beginning with Information 1. The PPK is also flanked by the common F and R primers, and contains in the proper order the unique Seq Primers for the ordered retrieval from each iDNA of its information segment sequence. Common spacer sequences indicate the demarcations between each Seq primer.

Each information segment should be capable of encoding any possible data (e.g., text), while correct readout requires that each Seq Primer prime a sequencing reaction only from the appropriate position within a specific iDNA, and not mis-prime on any iDNA. Various approaches can be taken to satisfy these conditions. In the simplest possible model, two bases (e.g., A and T) would be employed to encode text in information segments, and the other two (e.g., G and C) to construct Seq Primers. This would prevent mis-hybridization of Seq Primers to information segments, but would greatly limit both efficiency of text storage and the number of possible different Seq Primers.

To retrieve the stored information, a reader will amplify and sequence the PPK. To facilitate this step, the PPK plus the F and R PCR amplification primers can be stored separately from the iDNAs. The PPK, plus F and R primers and other non-degradable PCR reagents, could be stored in, or attached to, a vessel made of glass or plastic exhibiting long-term stability; and a separate vessel used to store the iDNAs. The outside of the former vessel could have a permanent glyph depicting PCR amplification, prompting a future reader to add a thermostable DNA polymerase (plus water) and begin PCR. This scheme should lead the reader, without further prompts, to sequence and interpret the PPK. Other long-term storage scenarios can be envisaged employing only a single container, involving, e.g., the use of DNA strands immobilized on beads. PCR amplification will yield levels of the PPK sufficient for further analysis, even if extensive degradation/modification had occurred during storage. Sequence analysis of the entire PPK reveals the sequences of the F and R PCR primers, plus an internal ordered series of elements of comparable size, suggesting roles for these elements as sequencing primers. This interpretation would lead the reader to perform sequence analysis of the information segments. Assuming that the F and R primers were interpreted to be "universal" PCR primers, the reader would employ these for simultaneous PCR amplification of all of the iDNAs. Sequential use of each Seq Primer to prime a sequencing reaction on the collection of PCR products would then yield the sequence of each of the information segments, arranged in the proper order to be decoded and read as a continuous block.

The two standard molecular biological techniques employed in the model for information storage in DNA could form the basis for a variety of DNA- based memory storage devices. The combined operations of PCR followed by sequence analysis are directly analogous to the retrieval of information from an addressable storage device such as the random access memory (RAM) in a computer. The ability to employ these combined operations to retrieve data permits construction of DNA representations of classical computer data structures such as arrays, linked lists, and trees. The model depicted in Fig. 1 is somewhat analogous to an array data structure. The PPK contains the addresses of the data elements (the iDNA information segments), which can be calculated (sequenced), and then employed for selective retrieval of the stored data. In an alternative serial model, analogous to a linked list, a series of iDNAs could be designed, each containing both a data element and the sequencing primer for retrieving the data element from the succeeding iDNA. Such a serial model would obviate the need for a separate PPK, but information retrieval would require considerably more experimental manipulations (and prior specific knowledge) than in the parallel model explored in detail here. Moreover, the use of DNA microarray technology should permit extensive scaleup of this model. Current microarray technology, in which up to 10,000 small DNA samples can be spotted in an ordered array onto a ~3 cm² surface (glass, etc.) (Gerhold, D. et al., 1999 TIBS 24:168) would have to be modified to permit spotting of DNA into small wells at a comparable density on a "microchip". Use of such a microchip for storage would impose two levels of order on the information stored in iDNAs placed in these "microwells": the X and Y coordinates of each micro well, plus the order within each microwell provided by the scheme described above and in Figure 1. A single series of unique identification primers, encoded within a single PPK (perhaps stored in the first microwell), should suffice to order the collection of iDNAs within every microwell, and thus permit readout in the proper order of the information stored on the entire chip. Because of the enormous number of different potential 20 base primer

■ 90 • ■ sequences, i.e., a theoretical maximum of 4 , the capacity for information storage in microarrayed DNA is presently limited by practical rather than theoretical considerations. It seems reasonable that with minor advances in microarray technology, about 700 novels or other data each equivalent in size to the entire "Tale of Two Cities" could be stored in a DNA microchip with the area of a postage stamp. A conservative upper limit on the size of the information segment is -600 bases, set by the present limits on DNA sequence obtainable from a single sequencing primer. If four-base codons chosen from our three-base alphabet (A, C, T) were employed (to pennit encoding of all common English alphanumeric characters plus a space), each iDNA could store about 150 characters. Storage of "A Tale of Two Cities", containing 135,345 words, or roughly 200,000 alphanumeric characters, would require -1300 iDNAs. Current technology would permit single- pass sequence analysis of a single PPK containing up to 100 unique sequencing primers, implying that information could be stored in and retrieved from about 100 different iDNAs per microwell. Since 14 microwells would thus be required to store Dickens' novel in DNA form, a 10,000 well microchip could store -700 texts or equivalent information the size of that novel.

6. EXAMPLE: DESIGN AND PRODUCTION OF INFORMATIONAL

DNA hi the example described herein the following method has been utilized. Text is encoded using only the bases A, C, and T. Seq Primers were designed using all four bases, plus a requirement that each 4^th position be a G. The resultant mismatch at (at least) each 4^th position between the sequences of any Seq Primer and any information segment should prevent mis-priming. Scaleup of the storage model presented here will ultimately require computer-generated design (see, e.g., M. Garzon, R. Deaton, J. A. Rose, in DNA Based Computers. DIMACS Series in Discrete Mathematics and Theoretical Computer Science vol. 54, E. Winfree,

D.K. Gifford, Eds. (American Mathematical Society, Providence, RI, 2000), p. 91) of large numbers of both Seq Primer and information segment DNA sequences satisfying the constraints on these elements.

A simple prototype of this technique was carried out, employing only the bases A, C, and T (i.e. omitting G) to encode text. To facilitate future decoding of the information stored in the iDNAs, an "obvious" ternary code based upon alphabetical order was employed to store English text in DNA. The DNA bases were ordered "alphabetically" (A, C, T). DNA codons were then constructed via a ternary code, beginning with "AAA" (encoding the letter "A"). The bases C, and then T, were inserted progressively into the 3^rd, 2^nd, and 1^st positions, yielding a series of 27 codons encoding the English letters in alphabetical order, plus a space. However, even if this encoding were not obvious to a future reader, recovery of a sufficient number of information segment sequences would permit use of standard cryptanalytical techniques- to determine how text had been encoded in the DNA. Two iDNAs were constructed to encode, respectively, "IT WAS THE

BEST OF TIMES IT WAS THE WORST OF TIMES", and "IT WAS THE AGE OF FOOLISHNESS IT WAS THE EPOCH OF BELIEF". Not only is this text one of the most famous opening lines of a novel C. Dickens, A Tale of Two Cities (Oxford University Press, London, New York, 1953. Originally published in 1859), but the four-fold repetition of the phrase "it was the" provided a test of the ability of this approach to deal with repeated DNA sequences both within and between information segments. The scheme described above was employed to simultaneously PCR amplify the two iDNAs contained in a microtube, and then to sequence each of the two amplification products. Decoding of the resultant DNA sequences above successfully recovered the stored text.

The two Information DNAs (iDNAs) employed for the prototype experiment were designed as described in the Letter. All primers were designed to contain a G in every 4th position, and also to contain sequences that would minimize the possibility of cross-hybridization among the primers. To ensure complete sequencing of the Information segment in each iDNA, a large 19-base common spacer sequence of alternating GTs was employed (substituting for the original design containing 19 straight Gs, which could not be synthesized by the supplier). However, since current sequencing technology permits determination of sequences quite close to the primer, it should be possible to employ instead small (3-4 base) common spacer sequences containing only Gs. The sequences of the common Forward and Reverse PCR primers (5* to 3'), are, respectively, TGCACGTCAGGAGGTAGGTC and TGCTCACTAGCGCACACGCT. The sequences of the unique Sequencing Primers for the first and second iDNAs (5' to 3') are, respectively,

TAGAGGGACTGTTCGGCAGC and CCGTTCGGAGGATAGCGAGT.

The two iDNAs (232 and 247 bases, respectively) were too long for commercial synthesis. They were therefore each constructed from overlapping short oligonucleotides (supplied by Genosys, Inc.), employing a modification of the single- step assembly PCR method (W.P.C. Stemmer, A. Crameri, K.D. Ha, T.M. Brennan, H.L. Heyneker, 1995, Gene 164, 49). These modifications, introduced to ensure specificity of formation of DNA molecules containing several repetitive regions, included use of larger starting oligonucleootides (23-104 bases), lower MgCl₂ concentration (1.5mM) in the gene assembly mix, and higher annealing temperatures (60°C). Following this use of assembly PCR to generate long, double stranded oligonucleotides corresponding to each iDNA, these oligonucleotides were PCR- amplified (25 cycles) as described (Ibid.), employing the common Forward and Reverse PCR Primer sequences described above. Following gel electrophoresis of the PCR products (on 3% low-melting agarose), fragments of the expected size were exercised and cloned into Pcr2.1-TOPO (Invitrogen), permitting storage of each iDNA in the form of a bacterial glycerol stock. To generate iDNAs for information storage, 50 ng of each cloned vector was employed as a template for PCR amplification. The PCR mix contained the common Forward and Reverse PCR Primers (final concentrations 0.5 uM), O.lmM dNTPs, 1U Taq polymerase plus standard buffers supplied by Qiagen. PCR involved an initial denaturation stage at 94°C for 5 min, followed by 30 cycles of 94°C for 30 s, 60°C for 30 s, 72°C for 45 s, and then a final extension step of 72°c for 5 min. Equal quantities (20 ng) of each iDNA were added to a microtube. The iDNAs were then simultaneously amplified in a PCR reaction identical to that described above, using the common Forward and Reverse PCR Primers. Aliquots (160 ng) of the resultant mixture of amplification products were employed for two sequencing reactions, each using 3 pmole of one of the above unique iDNA

Sequencing Primers, and the products analyzed on an ABI 377 sequencer. The Codon Table was then used to decode the resulting sequences of the two Information segments. This procedure returned the correct input text demonstrating the success of the prototype experiment.

Claims

We cla m

1. A method for storing information in a nucleic acid molecule comprising:

(i) synthesis of an information nucleic acid molecule wherein said information is stored as a nucleotide sequence;

(ii) synthesis of a polyprimer key wherein said polyprimer key provides information necessary for amplification and sequencing of the information nucleic acid molecule, thereby, permitting retrieval of the stored information.

2. The method of claim 1 wherein said information nucleic acid molecule comprises;

(i) a forward and reverse polymerase chain reaction primer recognition sequence;

(ii) a sequencing primer recognition sequence; (iii) at least one common spacer; and (iv) an information segment.

3. The method of claim 1 wherein said polyprimer key comprises:

(ii) at least one common spacer; and

(iii) a sequencing primer recognition sequence.

4. A method for retrieving stored information in a nucleic acid molecule comprising:

(i) amplification of a polyprimer key nucleic acid molecule using primers that bind to the primer recognition sequences of the polyprimer key;

(ii) deriving the nucleotide sequence of the sequencing primer recognition sequence, thereby, providing the primer sequence required for amplification and sequencing of an information nucleic acid molecule;

(iii) amplification of the information nucleic acid molecule using primers that bind to the primer recognition sequences of the information nucleic acid molecule; and

(iv) deriving the nucleotide sequence of the information nucleic acid molecule.

5. The method of claim 1 wherein the information nucleic acid molecule is stored on a microchip.

6. The method of claim 1 wherein the infonnation nucleic acid molecule is co- linear with the polyprimer key.